Paper/patches/server/0762-Highly-optimise-single-and-multi-AABB-VoxelShapes-an.patch

2121 lines
104 KiB
Diff
Raw Normal View History

Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
Rewrite chunk system (#8177) Patch documentation to come Issues with the old system that are fixed now: - World generation does not scale with cpu cores effectively. - Relies on the main thread for scheduling and maintaining chunk state, dropping chunk load/generate rates at lower tps. - Unreliable prioritisation of chunk gen/load calls that block the main thread. - Shutdown logic is utterly unreliable, as it has to wait for all chunks to unload - is it guaranteed that the chunk system is in a state on shutdown that it can reliably do this? Watchdog shutdown also typically failed due to thread checks, which is now resolved. - Saving of data is not unified (i.e can save chunk data without saving entity data, poses problems for desync if shutdown is really abnormal. - Entities are not loaded with chunks. This caused quite a bit of headache for Chunk#getEntities API, but now the new chunk system loads entities with chunks so that they are ready whenever the chunk loads in. Effectively brings the behavior back to 1.16 era, but still storing entities in their own separate regionfiles. The above list is not complete. The patch documentation will complete it. New chunk system hard relies on starlight and dataconverter, and most importantly the new concurrent utilities in ConcurrentUtil. Some of the old async chunk i/o interface (i.e the old file io thread reroutes _some_ calls to the new file io thread) is kept for plugin compat reasons. It will be removed in the next major version of minecraft. The old legacy chunk system patches have been moved to the removed folder in case we need them again.
2022-09-26 10:02:51 +02:00
From: Spottedleaf <Spottedleaf@users.noreply.github.com>
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
Date: Mon, 4 May 2020 10:06:24 -0700
Subject: [PATCH] Highly optimise single and multi-AABB VoxelShapes and
collisions
diff --git a/src/main/java/io/papermc/paper/util/CachedLists.java b/src/main/java/io/papermc/paper/util/CachedLists.java
index be668387f65a633c6ac497fca632a4767a1bf3a2..e08f4e39db4ee3fed62e37364d17dcc5c5683504 100644
--- a/src/main/java/io/papermc/paper/util/CachedLists.java
+++ b/src/main/java/io/papermc/paper/util/CachedLists.java
@@ -1,8 +1,57 @@
package io.papermc.paper.util;
+import net.minecraft.world.entity.Entity;
+import net.minecraft.world.phys.AABB;
+import org.bukkit.Bukkit;
+import org.bukkit.craftbukkit.util.UnsafeList;
+import java.util.List;
+
public final class CachedLists {
- public static void reset() {
+ // Paper start - optimise collisions
+ static final UnsafeList<AABB> TEMP_COLLISION_LIST = new UnsafeList<>(1024);
+ static boolean tempCollisionListInUse;
+
+ public static UnsafeList<AABB> getTempCollisionList() {
+ if (!Bukkit.isPrimaryThread() || tempCollisionListInUse) {
+ return new UnsafeList<>(16);
+ }
+ tempCollisionListInUse = true;
+ return TEMP_COLLISION_LIST;
+ }
+
+ public static void returnTempCollisionList(List<AABB> list) {
+ if (list != TEMP_COLLISION_LIST) {
+ return;
+ }
+ ((UnsafeList)list).setSize(0);
+ tempCollisionListInUse = false;
+ }
+ static final UnsafeList<Entity> TEMP_GET_ENTITIES_LIST = new UnsafeList<>(1024);
+ static boolean tempGetEntitiesListInUse;
+
+ public static UnsafeList<Entity> getTempGetEntitiesList() {
+ if (!Bukkit.isPrimaryThread() || tempGetEntitiesListInUse) {
+ return new UnsafeList<>(16);
+ }
+ tempGetEntitiesListInUse = true;
+ return TEMP_GET_ENTITIES_LIST;
+ }
+
+ public static void returnTempGetEntitiesList(List<Entity> list) {
+ if (list != TEMP_GET_ENTITIES_LIST) {
+ return;
+ }
+ ((UnsafeList)list).setSize(0);
+ tempGetEntitiesListInUse = false;
+ }
+ // Paper end - optimise collisions
+
+ public static void reset() {
+ // Paper start - optimise collisions
+ TEMP_COLLISION_LIST.completeReset();
+ TEMP_GET_ENTITIES_LIST.completeReset();
+ // Paper end - optimise collisions
}
}
diff --git a/src/main/java/io/papermc/paper/util/CollisionUtil.java b/src/main/java/io/papermc/paper/util/CollisionUtil.java
new file mode 100644
2022-02-28 21:34:23 +01:00
index 0000000000000000000000000000000000000000..a87f6380b2c387fb0cdd40d5087b5c93492e3c88
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- /dev/null
+++ b/src/main/java/io/papermc/paper/util/CollisionUtil.java
@@ -0,0 +1,899 @@
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+package io.papermc.paper.util;
+
+import io.papermc.paper.voxel.AABBVoxelShape;
+import net.minecraft.core.BlockPos;
+import net.minecraft.server.level.ServerChunkCache;
+import net.minecraft.server.level.ServerLevel;
+import net.minecraft.server.level.WorldGenRegion;
+import net.minecraft.util.Mth;
+import net.minecraft.world.entity.Entity;
+import net.minecraft.world.item.Item;
+import net.minecraft.world.level.CollisionGetter;
+import net.minecraft.world.level.EntityGetter;
+import net.minecraft.world.level.block.Blocks;
+import net.minecraft.world.level.block.state.BlockState;
+import net.minecraft.world.level.border.WorldBorder;
+import net.minecraft.world.level.chunk.ChunkAccess;
+import net.minecraft.world.level.chunk.LevelChunkSection;
+import net.minecraft.world.level.chunk.PalettedContainer;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+import net.minecraft.world.level.material.FlowingFluid;
+import net.minecraft.world.level.material.FluidState;
+import net.minecraft.world.phys.AABB;
+import net.minecraft.world.phys.Vec3;
+import net.minecraft.world.phys.shapes.ArrayVoxelShape;
+import net.minecraft.world.phys.shapes.CollisionContext;
+import net.minecraft.world.phys.shapes.EntityCollisionContext;
+import net.minecraft.world.phys.shapes.Shapes;
+import net.minecraft.world.phys.shapes.VoxelShape;
+import java.util.List;
+import java.util.function.BiPredicate;
+import java.util.function.Predicate;
+
+public final class CollisionUtil {
+
+ public static final double COLLISION_EPSILON = 1.0E-7;
+
+ public static final long KNOWN_EMPTY_BLOCK = 0b00; // known to always have voxelshape of empty
+ public static final long KNOWN_FULL_BLOCK = 0b01; // known to always have voxelshape of full cube
+ public static final long KNOWN_UNKNOWN_BLOCK = 0b10; // must read the actual block state for info
+ public static final long KNOWN_SPECIAL_BLOCK = 0b11; // caller must check this block for special collisions
+
+ public static boolean isSpecialCollidingBlock(final net.minecraft.world.level.block.state.BlockBehaviour.BlockStateBase block) {
+ return block.shapeExceedsCube() || block.getBlock() == Blocks.MOVING_PISTON;
+ }
+
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ public static boolean isEmpty(final AABB aabb) {
+ return (aabb.maxX - aabb.minX) < COLLISION_EPSILON && (aabb.maxY - aabb.minY) < COLLISION_EPSILON && (aabb.maxZ - aabb.minZ) < COLLISION_EPSILON;
+ }
+
+ public static boolean isEmpty(final double minX, final double minY, final double minZ,
+ final double maxX, final double maxY, final double maxZ) {
+ return (maxX - minX) < COLLISION_EPSILON && (maxY - minY) < COLLISION_EPSILON && (maxZ - minZ) < COLLISION_EPSILON;
+ }
+
+ public static AABB getBoxForChunk(final int chunkX, final int chunkZ) {
+ double x = (double)(chunkX << 4);
+ double z = (double)(chunkZ << 4);
+ // use a bounding box bigger than the chunk to prevent entities from entering it on move
+ return new AABB(x - 3*COLLISION_EPSILON, Double.NEGATIVE_INFINITY, z - 3*COLLISION_EPSILON,
+ x + (16.0 + 3*COLLISION_EPSILON), Double.POSITIVE_INFINITY, z + (16.0 + 3*COLLISION_EPSILON), false);
+ }
+
+ /*
+ A couple of rules for VoxelShape collisions:
+ Two shapes only intersect if they are actually more than EPSILON units into each other. This also applies to movement
+ checks.
+ If the two shapes strictly collide, then the return value of a collide call will return a value in the opposite
+ direction of the source move. However, this value will not be greater in magnitude than EPSILON. Collision code
+ will automatically round it to 0.
+ */
+
+ public static boolean voxelShapeIntersect(final double minX1, final double minY1, final double minZ1, final double maxX1,
+ final double maxY1, final double maxZ1, final double minX2, final double minY2,
+ final double minZ2, final double maxX2, final double maxY2, final double maxZ2) {
+ return (minX1 - maxX2) < -COLLISION_EPSILON && (maxX1 - minX2) > COLLISION_EPSILON &&
+ (minY1 - maxY2) < -COLLISION_EPSILON && (maxY1 - minY2) > COLLISION_EPSILON &&
+ (minZ1 - maxZ2) < -COLLISION_EPSILON && (maxZ1 - minZ2) > COLLISION_EPSILON;
+ }
+
+ public static boolean voxelShapeIntersect(final AABB box, final double minX, final double minY, final double minZ,
+ final double maxX, final double maxY, final double maxZ) {
+ return (box.minX - maxX) < -COLLISION_EPSILON && (box.maxX - minX) > COLLISION_EPSILON &&
+ (box.minY - maxY) < -COLLISION_EPSILON && (box.maxY - minY) > COLLISION_EPSILON &&
+ (box.minZ - maxZ) < -COLLISION_EPSILON && (box.maxZ - minZ) > COLLISION_EPSILON;
+ }
+
+ public static boolean voxelShapeIntersect(final AABB box1, final AABB box2) {
+ return (box1.minX - box2.maxX) < -COLLISION_EPSILON && (box1.maxX - box2.minX) > COLLISION_EPSILON &&
+ (box1.minY - box2.maxY) < -COLLISION_EPSILON && (box1.maxY - box2.minY) > COLLISION_EPSILON &&
+ (box1.minZ - box2.maxZ) < -COLLISION_EPSILON && (box1.maxZ - box2.minZ) > COLLISION_EPSILON;
+ }
+
+ public static double collideX(final AABB target, final AABB source, final double source_move) {
+ if (source_move == 0.0) {
+ return 0.0;
+ }
+
+ if ((source.minY - target.maxY) < -COLLISION_EPSILON && (source.maxY - target.minY) > COLLISION_EPSILON &&
+ (source.minZ - target.maxZ) < -COLLISION_EPSILON && (source.maxZ - target.minZ) > COLLISION_EPSILON) {
+ if (source_move >= 0.0) {
+ final double max_move = target.minX - source.maxX; // < 0.0 if no strict collision
+ if (max_move < -COLLISION_EPSILON) {
+ return source_move;
+ }
+ return Math.min(max_move, source_move);
+ } else {
+ final double max_move = target.maxX - source.minX; // > 0.0 if no strict collision
+ if (max_move > COLLISION_EPSILON) {
+ return source_move;
+ }
+ return Math.max(max_move, source_move);
+ }
+ }
+ return source_move;
+ }
+
+ public static double collideY(final AABB target, final AABB source, final double source_move) {
+ if (source_move == 0.0) {
+ return 0.0;
+ }
+
+ if ((source.minX - target.maxX) < -COLLISION_EPSILON && (source.maxX - target.minX) > COLLISION_EPSILON &&
+ (source.minZ - target.maxZ) < -COLLISION_EPSILON && (source.maxZ - target.minZ) > COLLISION_EPSILON) {
+ if (source_move >= 0.0) {
+ final double max_move = target.minY - source.maxY; // < 0.0 if no strict collision
+ if (max_move < -COLLISION_EPSILON) {
+ return source_move;
+ }
+ return Math.min(max_move, source_move);
+ } else {
+ final double max_move = target.maxY - source.minY; // > 0.0 if no strict collision
+ if (max_move > COLLISION_EPSILON) {
+ return source_move;
+ }
+ return Math.max(max_move, source_move);
+ }
+ }
+ return source_move;
+ }
+
+ public static double collideZ(final AABB target, final AABB source, final double source_move) {
+ if (source_move == 0.0) {
+ return 0.0;
+ }
+
+ if ((source.minX - target.maxX) < -COLLISION_EPSILON && (source.maxX - target.minX) > COLLISION_EPSILON &&
+ (source.minY - target.maxY) < -COLLISION_EPSILON && (source.maxY - target.minY) > COLLISION_EPSILON) {
+ if (source_move >= 0.0) {
+ final double max_move = target.minZ - source.maxZ; // < 0.0 if no strict collision
+ if (max_move < -COLLISION_EPSILON) {
+ return source_move;
+ }
+ return Math.min(max_move, source_move);
+ } else {
+ final double max_move = target.maxZ - source.minZ; // > 0.0 if no strict collision
+ if (max_move > COLLISION_EPSILON) {
+ return source_move;
+ }
+ return Math.max(max_move, source_move);
+ }
+ }
+ return source_move;
+ }
+
+ public static AABB offsetX(final AABB box, final double dx) {
+ return new AABB(box.minX + dx, box.minY, box.minZ, box.maxX + dx, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB offsetY(final AABB box, final double dy) {
+ return new AABB(box.minX, box.minY + dy, box.minZ, box.maxX, box.maxY + dy, box.maxZ, false);
+ }
+
+ public static AABB offsetZ(final AABB box, final double dz) {
+ return new AABB(box.minX, box.minY, box.minZ + dz, box.maxX, box.maxY, box.maxZ + dz, false);
+ }
+
+ public static AABB expandRight(final AABB box, final double dx) { // dx > 0.0
+ return new AABB(box.minX, box.minY, box.minZ, box.maxX + dx, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB expandLeft(final AABB box, final double dx) { // dx < 0.0
+ return new AABB(box.minX - dx, box.minY, box.minZ, box.maxX, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB expandUpwards(final AABB box, final double dy) { // dy > 0.0
+ return new AABB(box.minX, box.minY, box.minZ, box.maxX, box.maxY + dy, box.maxZ, false);
+ }
+
+ public static AABB expandDownwards(final AABB box, final double dy) { // dy < 0.0
+ return new AABB(box.minX, box.minY - dy, box.minZ, box.maxX, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB expandForwards(final AABB box, final double dz) { // dz > 0.0
+ return new AABB(box.minX, box.minY, box.minZ, box.maxX, box.maxY, box.maxZ + dz, false);
+ }
+
+ public static AABB expandBackwards(final AABB box, final double dz) { // dz < 0.0
+ return new AABB(box.minX, box.minY, box.minZ - dz, box.maxX, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB cutRight(final AABB box, final double dx) { // dx > 0.0
+ return new AABB(box.maxX, box.minY, box.minZ, box.maxX + dx, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB cutLeft(final AABB box, final double dx) { // dx < 0.0
+ return new AABB(box.minX + dx, box.minY, box.minZ, box.minX, box.maxY, box.maxZ, false);
+ }
+
+ public static AABB cutUpwards(final AABB box, final double dy) { // dy > 0.0
+ return new AABB(box.minX, box.maxY, box.minZ, box.maxX, box.maxY + dy, box.maxZ, false);
+ }
+
+ public static AABB cutDownwards(final AABB box, final double dy) { // dy < 0.0
+ return new AABB(box.minX, box.minY + dy, box.minZ, box.maxX, box.minY, box.maxZ, false);
+ }
+
+ public static AABB cutForwards(final AABB box, final double dz) { // dz > 0.0
+ return new AABB(box.minX, box.minY, box.maxZ, box.maxX, box.maxY, box.maxZ + dz, false);
+ }
+
+ public static AABB cutBackwards(final AABB box, final double dz) { // dz < 0.0
+ return new AABB(box.minX, box.minY, box.minZ + dz, box.maxX, box.maxY, box.minZ, false);
+ }
+
+ public static double performCollisionsX(final AABB currentBoundingBox, double value, final List<AABB> potentialCollisions) {
+ for (int i = 0, len = potentialCollisions.size(); i < len; ++i) {
+ final AABB target = potentialCollisions.get(i);
+ value = collideX(target, currentBoundingBox, value);
+ }
+
+ return value;
+ }
+
+ public static double performCollisionsY(final AABB currentBoundingBox, double value, final List<AABB> potentialCollisions) {
+ for (int i = 0, len = potentialCollisions.size(); i < len; ++i) {
+ final AABB target = potentialCollisions.get(i);
+ value = collideY(target, currentBoundingBox, value);
+ }
+
+ return value;
+ }
+
+ public static double performCollisionsZ(final AABB currentBoundingBox, double value, final List<AABB> potentialCollisions) {
+ for (int i = 0, len = potentialCollisions.size(); i < len; ++i) {
+ final AABB target = potentialCollisions.get(i);
+ value = collideZ(target, currentBoundingBox, value);
+ }
+
+ return value;
+ }
+
+ public static Vec3 performCollisions(final Vec3 moveVector, AABB axisalignedbb, final List<AABB> potentialCollisions) {
+ double x = moveVector.x;
+ double y = moveVector.y;
+ double z = moveVector.z;
+
+ if (y != 0.0) {
+ y = performCollisionsY(axisalignedbb, y, potentialCollisions);
+ if (y != 0.0) {
+ axisalignedbb = offsetY(axisalignedbb, y);
+ }
+ }
+
+ final boolean xSmaller = Math.abs(x) < Math.abs(z);
+
+ if (xSmaller && z != 0.0) {
+ z = performCollisionsZ(axisalignedbb, z, potentialCollisions);
+ if (z != 0.0) {
+ axisalignedbb = offsetZ(axisalignedbb, z);
+ }
+ }
+
+ if (x != 0.0) {
+ x = performCollisionsX(axisalignedbb, x, potentialCollisions);
+ if (!xSmaller && x != 0.0) {
+ axisalignedbb = offsetX(axisalignedbb, x);
+ }
+ }
+
+ if (!xSmaller && z != 0.0) {
+ z = performCollisionsZ(axisalignedbb, z, potentialCollisions);
+ }
+
+ return new Vec3(x, y, z);
+ }
+
+ public static boolean addBoxesToIfIntersects(final VoxelShape shape, final AABB aabb, final List<AABB> list) {
+ if (shape instanceof AABBVoxelShape) {
+ final AABBVoxelShape shapeCasted = (AABBVoxelShape)shape;
+ if (voxelShapeIntersect(shapeCasted.aabb, aabb) && !isEmpty(shapeCasted.aabb)) {
+ list.add(shapeCasted.aabb);
+ return true;
+ }
+ return false;
+ } else if (shape instanceof ArrayVoxelShape) {
+ final ArrayVoxelShape shapeCasted = (ArrayVoxelShape)shape;
+ // this can be optimised by checking an "overall shape" first, but not needed
+
+ final double offX = shapeCasted.getOffsetX();
+ final double offY = shapeCasted.getOffsetY();
+ final double offZ = shapeCasted.getOffsetZ();
+
+ boolean ret = false;
+
+ for (final AABB boundingBox : shapeCasted.getBoundingBoxesRepresentation()) {
+ final double minX, minY, minZ, maxX, maxY, maxZ;
+ if (voxelShapeIntersect(aabb, minX = boundingBox.minX + offX, minY = boundingBox.minY + offY, minZ = boundingBox.minZ + offZ,
+ maxX = boundingBox.maxX + offX, maxY = boundingBox.maxY + offY, maxZ = boundingBox.maxZ + offZ)
+ && !isEmpty(minX, minY, minZ, maxX, maxY, maxZ)) {
+ list.add(new AABB(minX, minY, minZ, maxX, maxY, maxZ, false));
+ ret = true;
+ }
+ }
+
+ return ret;
+ } else {
+ final List<AABB> boxes = shape.toAabbs();
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ boolean ret = false;
+
+ for (int i = 0, len = boxes.size(); i < len; ++i) {
+ final AABB box = boxes.get(i);
+ if (voxelShapeIntersect(box, aabb) && !isEmpty(box)) {
+ list.add(box);
+ ret = true;
+ }
+ }
+
+ return ret;
+ }
+ }
+
+ public static void addBoxesTo(final VoxelShape shape, final List<AABB> list) {
+ if (shape instanceof AABBVoxelShape) {
+ final AABBVoxelShape shapeCasted = (AABBVoxelShape)shape;
+ if (!isEmpty(shapeCasted.aabb)) {
+ list.add(shapeCasted.aabb);
+ }
+ } else if (shape instanceof ArrayVoxelShape) {
+ final ArrayVoxelShape shapeCasted = (ArrayVoxelShape)shape;
+
+ final double offX = shapeCasted.getOffsetX();
+ final double offY = shapeCasted.getOffsetY();
+ final double offZ = shapeCasted.getOffsetZ();
+
+ for (final AABB boundingBox : shapeCasted.getBoundingBoxesRepresentation()) {
+ final AABB box = boundingBox.move(offX, offY, offZ);
+ if (!isEmpty(box)) {
+ list.add(box);
+ }
+ }
+ } else {
+ final List<AABB> boxes = shape.toAabbs();
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ for (int i = 0, len = boxes.size(); i < len; ++i) {
+ final AABB box = boxes.get(i);
+ if (!isEmpty(box)) {
+ list.add(box);
+ }
+ }
+ }
+ }
+
+ public static boolean isAlmostCollidingOnBorder(final WorldBorder worldborder, final AABB boundingBox) {
+ return isAlmostCollidingOnBorder(worldborder, boundingBox.minX, boundingBox.maxX, boundingBox.minZ, boundingBox.maxZ);
+ }
+
+ public static boolean isAlmostCollidingOnBorder(final WorldBorder worldborder, final double boxMinX, final double boxMaxX,
+ final double boxMinZ, final double boxMaxZ) {
+ final double borderMinX = worldborder.getMinX(); // -X
+ final double borderMaxX = worldborder.getMaxX(); // +X
+
+ final double borderMinZ = worldborder.getMinZ(); // -Z
+ final double borderMaxZ = worldborder.getMaxZ(); // +Z
+
+ return
+ // Not intersecting if we're smaller
+ !voxelShapeIntersect(
+ boxMinX + COLLISION_EPSILON, Double.NEGATIVE_INFINITY, boxMinZ + COLLISION_EPSILON,
+ boxMaxX - COLLISION_EPSILON, Double.POSITIVE_INFINITY, boxMaxZ - COLLISION_EPSILON,
+ borderMinX, Double.NEGATIVE_INFINITY, borderMinZ, borderMaxX, Double.POSITIVE_INFINITY, borderMaxZ
+ )
+ &&
+
+ // Are intersecting if we're larger
+ voxelShapeIntersect(
+ boxMinX - COLLISION_EPSILON, Double.NEGATIVE_INFINITY, boxMinZ - COLLISION_EPSILON,
+ boxMaxX + COLLISION_EPSILON, Double.POSITIVE_INFINITY, boxMaxZ + COLLISION_EPSILON,
+ borderMinX, Double.NEGATIVE_INFINITY, borderMinZ, borderMaxX, Double.POSITIVE_INFINITY, borderMaxZ
+ );
+ }
+
+ public static boolean isCollidingWithBorderEdge(final WorldBorder worldborder, final AABB boundingBox) {
+ return isCollidingWithBorderEdge(worldborder, boundingBox.minX, boundingBox.maxX, boundingBox.minZ, boundingBox.maxZ);
+ }
+
+ public static boolean isCollidingWithBorderEdge(final WorldBorder worldborder, final double boxMinX, final double boxMaxX,
+ final double boxMinZ, final double boxMaxZ) {
+ final double borderMinX = worldborder.getMinX() + COLLISION_EPSILON; // -X
+ final double borderMaxX = worldborder.getMaxX() - COLLISION_EPSILON; // +X
+
+ final double borderMinZ = worldborder.getMinZ() + COLLISION_EPSILON; // -Z
+ final double borderMaxZ = worldborder.getMaxZ() - COLLISION_EPSILON; // +Z
+
+ return boxMinX < borderMinX || boxMaxX > borderMaxX || boxMinZ < borderMinZ || boxMaxZ > borderMaxZ;
+ }
+
+ public static boolean getCollisionsForBlocksOrWorldBorder(final CollisionGetter getter, final Entity entity, final AABB aabb,
+ final List<AABB> into, final boolean loadChunks, final boolean collidesWithUnloaded,
+ final boolean checkBorder, final boolean checkOnly, final BiPredicate<BlockState, BlockPos> predicate) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ boolean ret = false;
+
+ if (checkBorder) {
+ if (CollisionUtil.isAlmostCollidingOnBorder(getter.getWorldBorder(), aabb)) {
+ if (checkOnly) {
+ return true;
+ } else {
+ CollisionUtil.addBoxesTo(getter.getWorldBorder().getCollisionShape(), into);
+ ret = true;
+ }
+ }
+ }
+
+ final int minBlockX = Mth.floor(aabb.minX - COLLISION_EPSILON) - 1;
+ final int maxBlockX = Mth.floor(aabb.maxX + COLLISION_EPSILON) + 1;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int minBlockY = Mth.floor(aabb.minY - COLLISION_EPSILON) - 1;
+ final int maxBlockY = Mth.floor(aabb.maxY + COLLISION_EPSILON) + 1;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int minBlockZ = Mth.floor(aabb.minZ - COLLISION_EPSILON) - 1;
+ final int maxBlockZ = Mth.floor(aabb.maxZ + COLLISION_EPSILON) + 1;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int minSection = WorldUtil.getMinSection(getter);
+ final int maxSection = WorldUtil.getMaxSection(getter);
+ final int minBlock = minSection << 4;
+ final int maxBlock = (maxSection << 4) | 15;
+
+ final BlockPos.MutableBlockPos mutablePos = new BlockPos.MutableBlockPos();
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ CollisionContext collisionShape = null;
+
+ // special cases:
+ if (minBlockY > maxBlock || maxBlockY < minBlock) {
+ // no point in checking
+ return ret;
+ }
+
+ final int minYIterate = Math.max(minBlock, minBlockY);
+ final int maxYIterate = Math.min(maxBlock, maxBlockY);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int minChunkX = minBlockX >> 4;
+ final int maxChunkX = maxBlockX >> 4;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int minChunkY = minBlockY >> 4;
+ final int maxChunkY = maxBlockY >> 4;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int minChunkYIterate = minYIterate >> 4;
+ final int maxChunkYIterate = maxYIterate >> 4;
+
+ final int minChunkZ = minBlockZ >> 4;
+ final int maxChunkZ = maxBlockZ >> 4;
+
+ final ServerChunkCache chunkProvider;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ if (getter instanceof WorldGenRegion) {
+ chunkProvider = null;
+ } else if (getter instanceof ServerLevel) {
+ chunkProvider = ((ServerLevel)getter).getChunkSource();
+ } else {
+ chunkProvider = null;
+ }
+
+ for (int currChunkZ = minChunkZ; currChunkZ <= maxChunkZ; ++currChunkZ) {
+ final int minZ = currChunkZ == minChunkZ ? minBlockZ & 15 : 0; // coordinate in chunk
+ final int maxZ = currChunkZ == maxChunkZ ? maxBlockZ & 15 : 15; // coordinate in chunk
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ for (int currChunkX = minChunkX; currChunkX <= maxChunkX; ++currChunkX) {
+ final int minX = currChunkX == minChunkX ? minBlockX & 15 : 0; // coordinate in chunk
+ final int maxX = currChunkX == maxChunkX ? maxBlockX & 15 : 15; // coordinate in chunk
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int chunkXGlobalPos = currChunkX << 4;
+ final int chunkZGlobalPos = currChunkZ << 4;
+ final ChunkAccess chunk;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ if (chunkProvider == null) {
+ chunk = (ChunkAccess)getter.getChunkForCollisions(currChunkX, currChunkZ);
+ } else {
+ chunk = loadChunks ? chunkProvider.getChunk(currChunkX, currChunkZ, true) : chunkProvider.getChunkAtIfLoadedImmediately(currChunkX, currChunkZ);
+ }
+
+ if (chunk == null) {
+ if (collidesWithUnloaded) {
+ if (checkOnly) {
+ return true;
+ } else {
+ into.add(getBoxForChunk(currChunkX, currChunkZ));
+ ret = true;
+ }
+ }
+ continue;
+ }
+
+ final LevelChunkSection[] sections = chunk.getSections();
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ // bound y
+
+ for (int currChunkY = minChunkYIterate; currChunkY <= maxChunkYIterate; ++currChunkY) {
+ final LevelChunkSection section = sections[currChunkY - minSection];
+ if (section == null || section.hasOnlyAir()) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // empty
+ continue;
+ }
+ final PalettedContainer<BlockState> blocks = section.states;
+
+ final int minY = currChunkY == minChunkYIterate ? minYIterate & 15 : 0; // coordinate in chunk
+ final int maxY = currChunkY == maxChunkYIterate ? maxYIterate & 15 : 15; // coordinate in chunk
+ final int chunkYGlobalPos = currChunkY << 4;
+
+ final boolean sectionHasSpecial = section.hasSpecialCollidingBlocks();
+
+ final int minXIterate;
+ final int maxXIterate;
+ final int minZIterate;
+ final int maxZIterate;
+ final int minYIterateLocal;
+ final int maxYIterateLocal;
+
+ if (!sectionHasSpecial) {
+ minXIterate = currChunkX == minChunkX ? minX + 1 : minX;
+ maxXIterate = currChunkX == maxChunkX ? maxX - 1 : maxX;
+ minZIterate = currChunkZ == minChunkZ ? minZ + 1 : minZ;
+ maxZIterate = currChunkZ == maxChunkZ ? maxZ - 1 : maxZ;
+ minYIterateLocal = currChunkY == minChunkY ? minY + 1 : minY;
+ maxYIterateLocal = currChunkY == maxChunkY ? maxY - 1 : maxY;
+ if (minXIterate > maxXIterate || minZIterate > maxZIterate) {
+ continue;
+ }
+ } else {
+ minXIterate = minX;
+ maxXIterate = maxX;
+ minZIterate = minZ;
+ maxZIterate = maxZ;
+ minYIterateLocal = minY;
+ maxYIterateLocal = maxY;
+ }
+
+ for (int currY = minYIterateLocal; currY <= maxYIterateLocal; ++currY) {
+ long collisionForHorizontal = section.getKnownBlockInfoHorizontalRaw(currY, minZIterate & 15);
+ for (int currZ = minZIterate; currZ <= maxZIterate; ++currZ,
+ collisionForHorizontal = (currZ & 1) == 0 ? section.getKnownBlockInfoHorizontalRaw(currY, currZ & 15) : collisionForHorizontal) {
+ // From getKnownBlockInfoHorizontalRaw:
+ // important detail: this returns 32 values, one for localZ = localZ & (~1) and one for localZ = localZ | 1
+ // the even localZ is the lower 32 bits, the odd is the upper 32 bits
+ // We want to use a bitset to only iterate over non-empty blocks.
+ // We need to build a bitset mask to and out the other collisions we just don't care at all about
+ // First, we need to build a bitset from 0..n*2 where n is the number of blocks on the x axis
+ // It's important to note that the iterate values can be outside [0, 15], but if they are,
+ // then none of the x or z loops would meet their conditions. So we can assume they are never
+ // out of bounds here
+ final int xAxisBits = (maxXIterate - minXIterate + 1) << 1; // << 1 -> * 2 // Never > 32
+ long bitset = (1L << xAxisBits) - 1;
+ // Now we need to offset it by 32 bits if current Z is odd (lower 32 bits is 16 block infos for even z, upper is for odd)
+ int shift = (currZ & 1) << 5; // this will be a LEFT shift
+ // Now we need to offset shift so that the bitset first position is at minXIterate
+ shift += (minXIterate << 1); // 0th pos -> 0th bit, 1st pos -> 2nd bit, ...
+
+ // all done
+ bitset = bitset << shift;
+ if ((collisionForHorizontal & bitset) == 0L) {
+ // All empty
+ continue;
+ }
+ for (int currX = minXIterate; currX <= maxXIterate; ++currX) {
+ final int localBlockIndex = (currX) | (currZ << 4) | (currY << 8);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ final int blockInfo = (int) LevelChunkSection.getKnownBlockInfo(localBlockIndex, collisionForHorizontal);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ switch (blockInfo) {
+ case (int) CollisionUtil.KNOWN_EMPTY_BLOCK: {
+ continue;
+ }
+ case (int) CollisionUtil.KNOWN_FULL_BLOCK: {
+ double blockX = (double)(currX | chunkXGlobalPos);
+ double blockY = (double)(currY | chunkYGlobalPos);
+ double blockZ = (double)(currZ | chunkZGlobalPos);
+ final AABB blockBox = new AABB(
+ blockX, blockY, blockZ,
+ blockX + 1.0, blockY + 1.0, blockZ + 1.0,
+ true
+ );
+ if (predicate != null) {
+ if (!voxelShapeIntersect(aabb, blockBox)) {
+ continue;
+ }
+ // fall through to get the block for the predicate
+ } else {
+ if (voxelShapeIntersect(aabb, blockBox)) {
+ if (checkOnly) {
+ return true;
+ } else {
+ into.add(blockBox);
+ ret = true;
+ }
+ }
+ continue;
+ }
+ }
+ // default: fall through to standard logic
+ }
+
+ int blockX = currX | chunkXGlobalPos;
+ int blockY = currY | chunkYGlobalPos;
+ int blockZ = currZ | chunkZGlobalPos;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ int edgeCount = ((blockX == minBlockX || blockX == maxBlockX) ? 1 : 0) +
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ ((blockY == minBlockY || blockY == maxBlockY) ? 1 : 0) +
+ ((blockZ == minBlockZ || blockZ == maxBlockZ) ? 1 : 0);
+ if (edgeCount == 3) {
+ continue;
+ }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ BlockState blockData = blocks.get(localBlockIndex);
+
+ if ((edgeCount != 1 || blockData.shapeExceedsCube()) && (edgeCount != 2 || blockData.getBlock() == Blocks.MOVING_PISTON)) {
+ mutablePos.set(blockX, blockY, blockZ);
+ if (collisionShape == null) {
+ collisionShape = new LazyEntityCollisionContext(entity);
+ }
+ VoxelShape voxelshape2 = blockData.getCollisionShape(getter, mutablePos, collisionShape);
+ if (voxelshape2 != Shapes.empty()) {
+ VoxelShape voxelshape3 = voxelshape2.move((double)blockX, (double)blockY, (double)blockZ);
+
+ if (predicate != null && !predicate.test(blockData, mutablePos)) {
+ continue;
+ }
+
+ if (checkOnly) {
+ if (voxelshape3.intersects(aabb)) {
+ return true;
+ }
+ } else {
+ ret |= addBoxesToIfIntersects(voxelshape3, aabb, into);
+ }
+ }
+ }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ }
+ }
+ }
+ }
+ }
+
+ return ret;
+ }
+
+ public static boolean getCollisionsForBlocksOrWorldBorderReference(final CollisionGetter getter, final Entity entity, final AABB aabb,
+ final List<AABB> into, final boolean loadChunks, final boolean collidesWithUnloaded,
+ final boolean checkBorder, final boolean checkOnly, final BiPredicate<BlockState, BlockPos> predicate) {
+ boolean ret = false;
+
+ if (checkBorder) {
+ if (CollisionUtil.isAlmostCollidingOnBorder(getter.getWorldBorder(), aabb)) {
+ if (checkOnly) {
+ return true;
+ } else {
+ CollisionUtil.addBoxesTo(getter.getWorldBorder().getCollisionShape(), into);
+ ret = true;
+ }
+ }
+ }
+
+ final int minBlockX = Mth.floor(aabb.minX - COLLISION_EPSILON) - 1;
+ final int maxBlockX = Mth.floor(aabb.maxX + COLLISION_EPSILON) + 1;
+
+ final int minBlockY = Mth.floor(aabb.minY - COLLISION_EPSILON) - 1;
+ final int maxBlockY = Mth.floor(aabb.maxY + COLLISION_EPSILON) + 1;
+
+ final int minBlockZ = Mth.floor(aabb.minZ - COLLISION_EPSILON) - 1;
+ final int maxBlockZ = Mth.floor(aabb.maxZ + COLLISION_EPSILON) + 1;
+
+ final int minSection = WorldUtil.getMinSection(getter);
+ final int maxSection = WorldUtil.getMaxSection(getter);
+ final int minBlock = minSection << 4;
+ final int maxBlock = (maxSection << 4) | 15;
+
+ final BlockPos.MutableBlockPos mutablePos = new BlockPos.MutableBlockPos();
+ CollisionContext collisionShape = null;
+
+ // special cases:
+ if (minBlockY > maxBlock || maxBlockY < minBlock) {
+ // no point in checking
+ return ret;
+ }
+
+ final int minYIterate = Math.max(minBlock, minBlockY);
+ final int maxYIterate = Math.min(maxBlock, maxBlockY);
+
+ final int minChunkX = minBlockX >> 4;
+ final int maxChunkX = maxBlockX >> 4;
+
+ final int minChunkY = minBlockY >> 4;
+ final int maxChunkY = maxBlockY >> 4;
+
+ final int minChunkYIterate = minYIterate >> 4;
+ final int maxChunkYIterate = maxYIterate >> 4;
+
+ final int minChunkZ = minBlockZ >> 4;
+ final int maxChunkZ = maxBlockZ >> 4;
+
+ final ServerChunkCache chunkProvider;
+ if (getter instanceof WorldGenRegion) {
+ chunkProvider = null;
+ } else if (getter instanceof ServerLevel) {
+ chunkProvider = ((ServerLevel)getter).getChunkSource();
+ } else {
+ chunkProvider = null;
+ }
+
+ for (int currChunkZ = minChunkZ; currChunkZ <= maxChunkZ; ++currChunkZ) {
+ final int minZ = currChunkZ == minChunkZ ? minBlockZ & 15 : 0; // coordinate in chunk
+ final int maxZ = currChunkZ == maxChunkZ ? maxBlockZ & 15 : 15; // coordinate in chunk
+
+ for (int currChunkX = minChunkX; currChunkX <= maxChunkX; ++currChunkX) {
+ final int minX = currChunkX == minChunkX ? minBlockX & 15 : 0; // coordinate in chunk
+ final int maxX = currChunkX == maxChunkX ? maxBlockX & 15 : 15; // coordinate in chunk
+
+ final int chunkXGlobalPos = currChunkX << 4;
+ final int chunkZGlobalPos = currChunkZ << 4;
+ final ChunkAccess chunk;
+ if (chunkProvider == null) {
+ chunk = (ChunkAccess)getter.getChunkForCollisions(currChunkX, currChunkZ);
+ } else {
+ chunk = loadChunks ? chunkProvider.getChunk(currChunkX, currChunkZ, true) : chunkProvider.getChunkAtIfLoadedImmediately(currChunkX, currChunkZ);
+ }
+
+ if (chunk == null) {
+ if (collidesWithUnloaded) {
+ if (checkOnly) {
+ return true;
+ } else {
+ into.add(getBoxForChunk(currChunkX, currChunkZ));
+ ret = true;
+ }
+ }
+ continue;
+ }
+
+ final LevelChunkSection[] sections = chunk.getSections();
+
+ // bound y
+ for (int currChunkY = minChunkYIterate; currChunkY <= maxChunkYIterate; ++currChunkY) {
+ final LevelChunkSection section = sections[currChunkY - minSection];
+ if (section == null || section.hasOnlyAir()) {
+ // empty
+ continue;
+ }
+ final PalettedContainer<BlockState> blocks = section.states;
+
+ final int minY = currChunkY == minChunkYIterate ? minYIterate & 15 : 0; // coordinate in chunk
+ final int maxY = currChunkY == maxChunkYIterate ? maxYIterate & 15 : 15; // coordinate in chunk
+ final int chunkYGlobalPos = currChunkY << 4;
+
+ for (int currY = minY; currY <= maxY; ++currY) {
+ for (int currZ = minZ; currZ <= maxZ; ++currZ) {
+ for (int currX = minX; currX <= maxX; ++currX) {
+ int localBlockIndex = (currX) | (currZ << 4) | ((currY) << 8);
+ int blockX = currX | chunkXGlobalPos;
+ int blockY = currY | chunkYGlobalPos;
+ int blockZ = currZ | chunkZGlobalPos;
+
+ int edgeCount = ((blockX == minBlockX || blockX == maxBlockX) ? 1 : 0) +
+ ((blockY == minBlockY || blockY == maxBlockY) ? 1 : 0) +
+ ((blockZ == minBlockZ || blockZ == maxBlockZ) ? 1 : 0);
+ if (edgeCount == 3) {
+ continue;
+ }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ BlockState blockData = blocks.get(localBlockIndex);
+ if (blockData.getBlockCollisionBehavior() == CollisionUtil.KNOWN_EMPTY_BLOCK) {
+ continue;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ if ((edgeCount != 1 || blockData.shapeExceedsCube()) && (edgeCount != 2 || blockData.getBlock() == Blocks.MOVING_PISTON)) {
+ mutablePos.set(blockX, blockY, blockZ);
+ if (collisionShape == null) {
+ collisionShape = new LazyEntityCollisionContext(entity);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ VoxelShape voxelshape2 = blockData.getCollisionShape(getter, mutablePos, collisionShape);
+ if (voxelshape2 != Shapes.empty()) {
+ VoxelShape voxelshape3 = voxelshape2.move((double)blockX, (double)blockY, (double)blockZ);
+
+ if (predicate != null && !predicate.test(blockData, mutablePos)) {
+ continue;
+ }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ if (checkOnly) {
+ if (voxelshape3.intersects(aabb)) {
+ return true;
+ }
+ } else {
+ ret |= addBoxesToIfIntersects(voxelshape3, aabb, into);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+
+ return ret;
+ }
+
+ public static boolean getEntityHardCollisions(final CollisionGetter getter, final Entity entity, AABB aabb,
+ final List<AABB> into, final boolean checkOnly, final Predicate<Entity> predicate) {
+ if (isEmpty(aabb) || !(getter instanceof EntityGetter entityGetter)) {
+ return false;
+ }
+
+ boolean ret = false;
+
+ // to comply with vanilla intersection rules, expand by -epsilon so we only get stuff we definitely collide with.
+ // Vanilla for hard collisions has this backwards, and they expand by +epsilon but this causes terrible problems
+ // specifically with boat collisions.
+ aabb = aabb.inflate(-COLLISION_EPSILON, -COLLISION_EPSILON, -COLLISION_EPSILON);
+ final List<Entity> entities = CachedLists.getTempGetEntitiesList();
+ try {
+ if (entity != null && entity.hardCollides()) {
+ entityGetter.getEntities(entity, aabb, predicate, entities);
+ } else {
+ entityGetter.getHardCollidingEntities(entity, aabb, predicate, entities);
+ }
+
+ for (int i = 0, len = entities.size(); i < len; ++i) {
+ final Entity otherEntity = entities.get(i);
+
+ if ((entity == null && otherEntity.canBeCollidedWith()) || (entity != null && entity.canCollideWith(otherEntity))) {
+ if (checkOnly) {
+ return true;
+ } else {
+ into.add(otherEntity.getBoundingBox());
+ ret = true;
+ }
+ }
+ }
+ } finally {
+ CachedLists.returnTempGetEntitiesList(entities);
+ }
+
+ return ret;
+ }
+
+ public static boolean getCollisions(final CollisionGetter view, final Entity entity, final AABB aabb,
+ final List<AABB> into, final boolean loadChunks, final boolean collidesWithUnloadedChunks,
+ final boolean checkBorder, final boolean checkOnly, final BiPredicate<BlockState, BlockPos> blockPredicate,
+ final Predicate<Entity> entityPredicate) {
+ if (checkOnly) {
+ return getCollisionsForBlocksOrWorldBorder(view, entity, aabb, into, loadChunks, collidesWithUnloadedChunks, checkBorder, checkOnly, blockPredicate)
+ || getEntityHardCollisions(view, entity, aabb, into, checkOnly, entityPredicate);
+ } else {
+ return getCollisionsForBlocksOrWorldBorder(view, entity, aabb, into, loadChunks, collidesWithUnloadedChunks, checkBorder, checkOnly, blockPredicate)
+ | getEntityHardCollisions(view, entity, aabb, into, checkOnly, entityPredicate);
+ }
+ }
+
+ public static final class LazyEntityCollisionContext extends EntityCollisionContext {
+
+ private CollisionContext delegate;
+
+ public LazyEntityCollisionContext(final Entity entity) {
+ super(false, 0.0, null, null, entity);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ public CollisionContext getDelegate() {
+ final Entity entity = this.getEntity();
+ return this.delegate == null ? this.delegate = (entity == null ? CollisionContext.empty() : CollisionContext.of(entity)) : this.delegate;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ @Override
+ public boolean isDescending() {
+ return this.getDelegate().isDescending();
+ }
+
+ @Override
+ public boolean isAbove(final VoxelShape shape, final BlockPos pos, final boolean defaultValue) {
+ return this.getDelegate().isAbove(shape, pos, defaultValue);
+ }
+
+ @Override
+ public boolean isHoldingItem(final Item item) {
+ return this.getDelegate().isHoldingItem(item);
+ }
+
+ @Override
2022-02-28 21:34:23 +01:00
+ public boolean canStandOnFluid(final FluidState state, final FluidState fluidState) {
+ return this.getDelegate().canStandOnFluid(state, fluidState);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ }
+
+ private CollisionUtil() {
+ throw new RuntimeException();
+ }
+}
diff --git a/src/main/java/io/papermc/paper/voxel/AABBVoxelShape.java b/src/main/java/io/papermc/paper/voxel/AABBVoxelShape.java
new file mode 100644
index 0000000000000000000000000000000000000000..d67a40e7be030142443680c89e1763fc9ecdfe0a
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- /dev/null
+++ b/src/main/java/io/papermc/paper/voxel/AABBVoxelShape.java
@@ -0,0 +1,200 @@
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+package io.papermc.paper.voxel;
+
+import io.papermc.paper.util.CollisionUtil;
+import it.unimi.dsi.fastutil.doubles.DoubleArrayList;
+import it.unimi.dsi.fastutil.doubles.DoubleList;
+import net.minecraft.core.Direction;
+import net.minecraft.world.phys.AABB;
+import net.minecraft.world.phys.shapes.Shapes;
+import net.minecraft.world.phys.shapes.VoxelShape;
+import java.util.ArrayList;
+import java.util.List;
+
+public final class AABBVoxelShape extends VoxelShape {
+
+ public final AABB aabb;
+
+ public AABBVoxelShape(AABB aabb) {
+ super(Shapes.getFullUnoptimisedCube().shape);
+ this.aabb = aabb;
+ }
+
+ @Override
+ public boolean isEmpty() {
+ return CollisionUtil.isEmpty(this.aabb);
+ }
+
+ @Override
+ public double min(Direction.Axis enumdirection_enumaxis) {
+ switch (enumdirection_enumaxis.ordinal()) {
+ case 0:
+ return this.aabb.minX;
+ case 1:
+ return this.aabb.minY;
+ case 2:
+ return this.aabb.minZ;
+ default:
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+
+ @Override
+ public double max(Direction.Axis enumdirection_enumaxis) {
+ switch (enumdirection_enumaxis.ordinal()) {
+ case 0:
+ return this.aabb.maxX;
+ case 1:
+ return this.aabb.maxY;
+ case 2:
+ return this.aabb.maxZ;
+ default:
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+
+ @Override
+ public AABB bounds() {
+ return this.aabb;
+ }
+
+ // enum direction axis is from 0 -> 2, so we keep the lower bits for direction axis.
+ @Override
+ protected double get(Direction.Axis enumdirection_enumaxis, int i) {
+ switch (enumdirection_enumaxis.ordinal() | (i << 2)) {
+ case (0 | (0 << 2)):
+ return this.aabb.minX;
+ case (1 | (0 << 2)):
+ return this.aabb.minY;
+ case (2 | (0 << 2)):
+ return this.aabb.minZ;
+ case (0 | (1 << 2)):
+ return this.aabb.maxX;
+ case (1 | (1 << 2)):
+ return this.aabb.maxY;
+ case (2 | (1 << 2)):
+ return this.aabb.maxZ;
+ default:
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+
+ private DoubleList cachedListX;
+ private DoubleList cachedListY;
+ private DoubleList cachedListZ;
+
+ @Override
+ protected DoubleList getCoords(Direction.Axis enumdirection_enumaxis) {
+ switch (enumdirection_enumaxis.ordinal()) {
+ case 0:
+ return this.cachedListX == null ? this.cachedListX = DoubleArrayList.wrap(new double[] { this.aabb.minX, this.aabb.maxX }) : this.cachedListX;
+ case 1:
+ return this.cachedListY == null ? this.cachedListY = DoubleArrayList.wrap(new double[] { this.aabb.minY, this.aabb.maxY }) : this.cachedListY;
+ case 2:
+ return this.cachedListZ == null ? this.cachedListZ = DoubleArrayList.wrap(new double[] { this.aabb.minZ, this.aabb.maxZ }) : this.cachedListZ;
+ default:
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+
+ @Override
+ public VoxelShape move(double d0, double d1, double d2) {
+ return new AABBVoxelShape(this.aabb.move(d0, d1, d2));
+ }
+
+ @Override
+ public VoxelShape optimize() {
+ if (this.isEmpty()) {
+ return Shapes.empty();
+ } else if (this == Shapes.BLOCK_OPTIMISED || this.aabb.equals(Shapes.BLOCK_OPTIMISED.aabb)) {
+ return Shapes.BLOCK_OPTIMISED;
+ }
+ return this;
+ }
+
+ @Override
+ public void forAllBoxes(Shapes.DoubleLineConsumer voxelshapes_a) {
+ voxelshapes_a.consume(this.aabb.minX, this.aabb.minY, this.aabb.minZ, this.aabb.maxX, this.aabb.maxY, this.aabb.maxZ);
+ }
+
+ @Override
+ public List<AABB> toAabbs() { // getAABBs
+ List<AABB> ret = new ArrayList<>(1);
+ ret.add(this.aabb);
+ return ret;
+ }
+
+ @Override
+ protected int findIndex(Direction.Axis enumdirection_enumaxis, double d0) { // findPointIndexAfterOffset
+ switch (enumdirection_enumaxis.ordinal()) {
+ case 0:
+ return d0 < this.aabb.maxX ? (d0 < this.aabb.minX ? -1 : 0) : 1;
+ case 1:
+ return d0 < this.aabb.maxY ? (d0 < this.aabb.minY ? -1 : 0) : 1;
+ case 2:
+ return d0 < this.aabb.maxZ ? (d0 < this.aabb.minZ ? -1 : 0) : 1;
+ default:
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+
+ @Override
+ protected VoxelShape calculateFace(Direction direction) {
+ if (this.isEmpty()) {
+ return Shapes.empty();
+ }
+ if (this == Shapes.BLOCK_OPTIMISED) {
+ return this;
+ }
+ switch (direction) {
+ case EAST: // +X
+ case WEST: { // -X
+ final double from = direction == Direction.EAST ? 1.0 - CollisionUtil.COLLISION_EPSILON : CollisionUtil.COLLISION_EPSILON;
+ if (from > this.aabb.maxX || this.aabb.minX > from) {
+ return Shapes.empty();
+ }
+ return new AABBVoxelShape(new AABB(0.0, this.aabb.minY, this.aabb.minZ, 1.0, this.aabb.maxY, this.aabb.maxZ)).optimize();
+ }
+ case UP: // +Y
+ case DOWN: { // -Y
+ final double from = direction == Direction.UP ? 1.0 - CollisionUtil.COLLISION_EPSILON : CollisionUtil.COLLISION_EPSILON;
+ if (from > this.aabb.maxY || this.aabb.minY > from) {
+ return Shapes.empty();
+ }
+ return new AABBVoxelShape(new AABB(this.aabb.minX, 0.0, this.aabb.minZ, this.aabb.maxX, 1.0, this.aabb.maxZ)).optimize();
+ }
+ case SOUTH: // +Z
+ case NORTH: { // -Z
+ final double from = direction == Direction.SOUTH ? 1.0 - CollisionUtil.COLLISION_EPSILON : CollisionUtil.COLLISION_EPSILON;
+ if (from > this.aabb.maxZ || this.aabb.minZ > from) {
+ return Shapes.empty();
+ }
+ return new AABBVoxelShape(new AABB(this.aabb.minX, this.aabb.minY, 0.0, this.aabb.maxX, this.aabb.maxY, 1.0)).optimize();
+ }
+ default: {
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+ }
+
+ @Override
+ public double collide(Direction.Axis enumdirection_enumaxis, AABB axisalignedbb, double d0) {
+ if (CollisionUtil.isEmpty(this.aabb) || CollisionUtil.isEmpty(axisalignedbb)) {
+ return d0;
+ }
+ switch (enumdirection_enumaxis.ordinal()) {
+ case 0:
+ return CollisionUtil.collideX(this.aabb, axisalignedbb, d0);
+ case 1:
+ return CollisionUtil.collideY(this.aabb, axisalignedbb, d0);
+ case 2:
+ return CollisionUtil.collideZ(this.aabb, axisalignedbb, d0);
+ default:
+ throw new IllegalStateException("Unknown axis requested");
+ }
+ }
+
+ @Override
+ public boolean intersects(AABB axisalingedbb) {
+ return CollisionUtil.voxelShapeIntersect(this.aabb, axisalingedbb);
+ }
+}
diff --git a/src/main/java/net/minecraft/server/level/ServerPlayer.java b/src/main/java/net/minecraft/server/level/ServerPlayer.java
index e1b59a1e6fd957b45f0a73eb51487e44108af34c..2e3d5b38ace99ce99a8107e750c5eeaf281c8552 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/server/level/ServerPlayer.java
+++ b/src/main/java/net/minecraft/server/level/ServerPlayer.java
@@ -411,7 +411,7 @@ public class ServerPlayer extends Player {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
if (blockposition1 != null) {
this.moveTo(blockposition1, 0.0F, 0.0F);
- if (world.noCollision((Entity) this)) {
+ if (world.noCollision(this, this.getBoundingBox(), true)) { // Paper - make sure this loads chunks, we default to NOT loading now
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
break;
}
}
@@ -419,7 +419,7 @@ public class ServerPlayer extends Player {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
} else {
this.moveTo(blockposition, 0.0F, 0.0F);
- while (!world.noCollision((Entity) this) && this.getY() < (double) (world.getMaxBuildHeight() - 1)) {
+ while (!world.noCollision(this, this.getBoundingBox(), true) && this.getY() < (double) (world.getMaxBuildHeight() - 1)) { // Paper - make sure this loads chunks, we default to NOT loading now
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
this.setPos(this.getX(), this.getY() + 1.0D, this.getZ());
}
}
diff --git a/src/main/java/net/minecraft/server/players/PlayerList.java b/src/main/java/net/minecraft/server/players/PlayerList.java
index 6c7d68f910d3be55ed0b3124bfc861d5446c1388..88e9c5faca16eeb5cafcedb811a3d9d4dba6e9ce 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/server/players/PlayerList.java
+++ b/src/main/java/net/minecraft/server/players/PlayerList.java
@@ -874,7 +874,7 @@ public abstract class PlayerList {
// CraftBukkit end
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
worldserver1.getChunkSource().addRegionTicket(net.minecraft.server.level.TicketType.POST_TELEPORT, new net.minecraft.world.level.ChunkPos(location.getBlockX() >> 4, location.getBlockZ() >> 4), 1, entityplayer.getId()); // Paper
2022-03-01 06:43:03 +01:00
- while (avoidSuffocation && !worldserver1.noCollision((Entity) entityplayer1) && entityplayer1.getY() < (double) worldserver1.getMaxBuildHeight()) {
+ while (avoidSuffocation && !worldserver1.noCollision(entityplayer1, entityplayer1.getBoundingBox(), true) && entityplayer1.getY() < (double) worldserver1.getMaxBuildHeight()) { // Paper - make sure this loads chunks, we default to NOT loading now
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
entityplayer1.setPos(entityplayer1.getX(), entityplayer1.getY() + 1.0D, entityplayer1.getZ());
}
2022-12-07 22:35:34 +01:00
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
diff --git a/src/main/java/net/minecraft/world/entity/Entity.java b/src/main/java/net/minecraft/world/entity/Entity.java
2022-12-23 20:59:11 +01:00
index b284a9fd506f30256b8194603c12865086eebdaa..48606a6d6ee4e1ba9d6cdb1600bbe8e62a16e400 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/entity/Entity.java
+++ b/src/main/java/net/minecraft/world/entity/Entity.java
2022-10-11 14:43:10 +02:00
@@ -1160,9 +1160,44 @@ public abstract class Entity implements Nameable, EntityAccess, CommandSource {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
float f2 = this.getBlockSpeedFactor();
this.setDeltaMovement(this.getDeltaMovement().multiply((double) f2, 1.0D, (double) f2));
- if (this.level.getBlockStatesIfLoaded(this.getBoundingBox().deflate(1.0E-6D)).noneMatch((iblockdata1) -> {
2022-03-01 06:43:03 +01:00
- return iblockdata1.is(BlockTags.FIRE) || iblockdata1.is(Blocks.LAVA);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
- })) {
+ // Paper start - remove expensive streams from here
+ boolean noneMatch = true;
+ AABB fireSearchBox = this.getBoundingBox().deflate(1.0E-6D);
+ {
+ int minX = Mth.floor(fireSearchBox.minX);
+ int minY = Mth.floor(fireSearchBox.minY);
+ int minZ = Mth.floor(fireSearchBox.minZ);
+ int maxX = Mth.floor(fireSearchBox.maxX);
+ int maxY = Mth.floor(fireSearchBox.maxY);
+ int maxZ = Mth.floor(fireSearchBox.maxZ);
+ fire_search_loop:
+ for (int fz = minZ; fz <= maxZ; ++fz) {
+ for (int fx = minX; fx <= maxX; ++fx) {
+ for (int fy = minY; fy <= maxY; ++fy) {
+ net.minecraft.world.level.chunk.LevelChunk chunk = (net.minecraft.world.level.chunk.LevelChunk)this.level.getChunkIfLoadedImmediately(fx >> 4, fz >> 4);
+ if (chunk == null) {
+ // Vanilla rets an empty stream if all the chunks are not loaded, so noneMatch will be true
+ // even if we're in lava/fire
+ noneMatch = true;
+ break fire_search_loop;
+ }
+ if (!noneMatch) {
+ // don't do get type, we already know we're in fire - we just need to check the chunks
+ // loaded state
+ continue;
+ }
+
+ BlockState type = chunk.getBlockStateFinal(fx, fy, fz);
2022-03-01 06:43:03 +01:00
+ if (type.is(BlockTags.FIRE) || type.is(Blocks.LAVA)) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ noneMatch = false;
+ // can't break, we need to retain vanilla behavior by ensuring ALL chunks are loaded
+ }
+ }
+ }
+ }
+ }
+ if (noneMatch) {
+ // Paper end - remove expensive streams from here
if (this.remainingFireTicks <= 0) {
this.setRemainingFireTicks(-this.getFireImmuneTicks());
}
2022-12-07 22:35:34 +01:00
@@ -1314,32 +1349,78 @@ public abstract class Entity implements Nameable, EntityAccess, CommandSource {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
private Vec3 collide(Vec3 movement) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
- AABB axisalignedbb = this.getBoundingBox();
- List<VoxelShape> list = this.level.getEntityCollisions(this, axisalignedbb.expandTowards(movement));
- Vec3 vec3d1 = movement.lengthSqr() == 0.0D ? movement : Entity.collideBoundingBox(this, movement, axisalignedbb, this.level, list);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
- boolean flag = movement.x != vec3d1.x;
- boolean flag1 = movement.y != vec3d1.y;
- boolean flag2 = movement.z != vec3d1.z;
- boolean flag3 = this.onGround || flag1 && movement.y < 0.0D;
-
- if (this.maxUpStep > 0.0F && flag3 && (flag || flag2)) {
- Vec3 vec3d2 = Entity.collideBoundingBox(this, new Vec3(movement.x, (double) this.maxUpStep, movement.z), axisalignedbb, this.level, list);
- Vec3 vec3d3 = Entity.collideBoundingBox(this, new Vec3(0.0D, (double) this.maxUpStep, 0.0D), axisalignedbb.expandTowards(movement.x, 0.0D, movement.z), this.level, list);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
-
- if (vec3d3.y < (double) this.maxUpStep) {
- Vec3 vec3d4 = Entity.collideBoundingBox(this, new Vec3(movement.x, 0.0D, movement.z), axisalignedbb.move(vec3d3), this.level, list).add(vec3d3);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
-
- if (vec3d4.horizontalDistanceSqr() > vec3d2.horizontalDistanceSqr()) {
- vec3d2 = vec3d4;
+ // Paper start - optimise collisions
+ // This is a copy of vanilla's except that it uses strictly AABB math
+ if (movement.x == 0.0 && movement.y == 0.0 && movement.z == 0.0) {
+ return movement;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ final Level world = this.level;
+ final AABB currBoundingBox = this.getBoundingBox();
+
+ if (io.papermc.paper.util.CollisionUtil.isEmpty(currBoundingBox)) {
+ return movement;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ final List<AABB> potentialCollisions = io.papermc.paper.util.CachedLists.getTempCollisionList();
+ try {
+ final double stepHeight = (double)this.maxUpStep;
+ final AABB collisionBox;
+
+ if (movement.x == 0.0 && movement.z == 0.0 && movement.y != 0.0) {
+ if (movement.y > 0.0) {
+ collisionBox = io.papermc.paper.util.CollisionUtil.cutUpwards(currBoundingBox, movement.y);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ } else {
+ collisionBox = io.papermc.paper.util.CollisionUtil.cutDownwards(currBoundingBox, movement.y);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ } else {
+ if (stepHeight > 0.0 && (this.onGround || (movement.y < 0.0)) && (movement.x != 0.0 || movement.z != 0.0)) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // don't bother getting the collisions if we don't need them.
+ if (movement.y <= 0.0) {
+ collisionBox = io.papermc.paper.util.CollisionUtil.expandUpwards(currBoundingBox.expandTowards(movement.x, movement.y, movement.z), stepHeight);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ } else {
+ collisionBox = currBoundingBox.expandTowards(movement.x, Math.max(stepHeight, movement.y), movement.z);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ } else {
+ collisionBox = currBoundingBox.expandTowards(movement.x, movement.y, movement.z);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
}
- if (vec3d2.horizontalDistanceSqr() > vec3d1.horizontalDistanceSqr()) {
- return vec3d2.add(Entity.collideBoundingBox(this, new Vec3(0.0D, -vec3d2.y + movement.y, 0.0D), axisalignedbb.move(vec3d2), this.level, list));
+ io.papermc.paper.util.CollisionUtil.getCollisions(world, this, collisionBox, potentialCollisions, false, this.level.paperConfig().chunks.preventMovingIntoUnloadedChunks,
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ false, false, null, null);
+
+ if (io.papermc.paper.util.CollisionUtil.isCollidingWithBorderEdge(world.getWorldBorder(), collisionBox)) {
+ io.papermc.paper.util.CollisionUtil.addBoxesToIfIntersects(world.getWorldBorder().getCollisionShape(), collisionBox, potentialCollisions);
}
- }
- return vec3d1;
+ final Vec3 limitedMoveVector = io.papermc.paper.util.CollisionUtil.performCollisions(movement, currBoundingBox, potentialCollisions);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ if (stepHeight > 0.0
+ && (this.onGround || (limitedMoveVector.y != movement.y && movement.y < 0.0))
+ && (limitedMoveVector.x != movement.x || limitedMoveVector.z != movement.z)) {
+ Vec3 vec3d2 = io.papermc.paper.util.CollisionUtil.performCollisions(new Vec3(movement.x, stepHeight, movement.z), currBoundingBox, potentialCollisions);
+ final Vec3 vec3d3 = io.papermc.paper.util.CollisionUtil.performCollisions(new Vec3(0.0, stepHeight, 0.0), currBoundingBox.expandTowards(movement.x, 0.0, movement.z), potentialCollisions);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ if (vec3d3.y < stepHeight) {
+ final Vec3 vec3d4 = io.papermc.paper.util.CollisionUtil.performCollisions(new Vec3(movement.x, 0.0D, movement.z), currBoundingBox.move(vec3d3), potentialCollisions).add(vec3d3);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+
+ if (vec3d4.horizontalDistanceSqr() > vec3d2.horizontalDistanceSqr()) {
+ vec3d2 = vec3d4;
+ }
+ }
+
+ if (vec3d2.horizontalDistanceSqr() > limitedMoveVector.horizontalDistanceSqr()) {
+ return vec3d2.add(io.papermc.paper.util.CollisionUtil.performCollisions(new Vec3(0.0D, -vec3d2.y + movement.y, 0.0D), currBoundingBox.move(vec3d2), potentialCollisions));
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ return limitedMoveVector;
+ } else {
+ return limitedMoveVector;
+ }
+ } finally {
+ io.papermc.paper.util.CachedLists.returnTempCollisionList(potentialCollisions);
+ }
+ // Paper end - optimise collisions
}
public static Vec3 collideBoundingBox(@Nullable Entity entity, Vec3 movement, AABB entityBoundingBox, Level world, List<VoxelShape> collisions) {
2022-12-23 20:59:11 +01:00
@@ -2478,11 +2559,30 @@ public abstract class Entity implements Nameable, EntityAccess, CommandSource {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
float f = this.dimensions.width * 0.8F;
2022-03-01 06:43:03 +01:00
AABB axisalignedbb = AABB.ofSize(this.getEyePosition(), (double) f, 1.0E-6D, (double) f);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
2022-03-01 06:43:03 +01:00
- return BlockPos.betweenClosedStream(axisalignedbb).anyMatch((blockposition) -> {
- BlockState iblockdata = this.level.getBlockState(blockposition);
+ BlockPos.MutableBlockPos blockposition = new BlockPos.MutableBlockPos();
+ int minX = Mth.floor(axisalignedbb.minX);
+ int minY = Mth.floor(axisalignedbb.minY);
+ int minZ = Mth.floor(axisalignedbb.minZ);
+ int maxX = Mth.floor(axisalignedbb.maxX);
+ int maxY = Mth.floor(axisalignedbb.maxY);
+ int maxZ = Mth.floor(axisalignedbb.maxZ);
+ for (int fz = minZ; fz <= maxZ; ++fz) {
+ for (int fx = minX; fx <= maxX; ++fx) {
+ for (int fy = minY; fy <= maxY; ++fy) {
+ net.minecraft.world.level.chunk.LevelChunk chunk = (net.minecraft.world.level.chunk.LevelChunk)this.level.getChunkIfLoadedImmediately(fx >> 4, fz >> 4);
+ if (chunk == null) {
+ continue;
+ }
- return !iblockdata.isAir() && iblockdata.isSuffocating(this.level, blockposition) && Shapes.joinIsNotEmpty(iblockdata.getCollisionShape(this.level, blockposition).move((double) blockposition.getX(), (double) blockposition.getY(), (double) blockposition.getZ()), Shapes.create(axisalignedbb), BooleanOp.AND);
- });
2022-03-01 06:43:03 +01:00
+ BlockState iblockdata = chunk.getBlockStateFinal(fx, fy, fz);
+ blockposition.set(fx, fy, fz);
+ if (!iblockdata.isAir() && iblockdata.isSuffocating(this.level, blockposition) && Shapes.joinIsNotEmpty(iblockdata.getCollisionShape(this.level, blockposition).move((double) blockposition.getX(), (double) blockposition.getY(), (double) blockposition.getZ()), Shapes.create(axisalignedbb), BooleanOp.AND)) {
+ return true;
+ }
+ }
+ }
+ }
+ return false;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
}
diff --git a/src/main/java/net/minecraft/world/level/BlockCollisions.java b/src/main/java/net/minecraft/world/level/BlockCollisions.java
2022-11-03 22:03:31 +01:00
index be578f14146b0184d5419d5b961c5d681f9ba7a3..018cbf866bbe39b69a4afa039166e8d34ec3ab5f 100644
--- a/src/main/java/net/minecraft/world/level/BlockCollisions.java
+++ b/src/main/java/net/minecraft/world/level/BlockCollisions.java
@@ -106,7 +106,7 @@ public class BlockCollisions extends AbstractIterator<VoxelShape> {
VoxelShape voxelShape = blockState.getCollisionShape(this.collisionGetter, this.pos, this.context);
if (voxelShape == Shapes.block()) {
- if (!this.box.intersects((double)i, (double)j, (double)k, (double)i + 1.0D, (double)j + 1.0D, (double)k + 1.0D)) {
+ if (!io.papermc.paper.util.CollisionUtil.voxelShapeIntersect(this.box, (double)i, (double)j, (double)k, (double)i + 1.0D, (double)j + 1.0D, (double)k + 1.0D)) { // Paper - keep vanilla behavior for voxelshape intersection - See comment in CollisionUtil
continue;
}
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
diff --git a/src/main/java/net/minecraft/world/level/CollisionGetter.java b/src/main/java/net/minecraft/world/level/CollisionGetter.java
index 56d94c94fb0d4dc468bb5d69be655ddd5c6b5360..d7d396ad73866a97cd9f63b34ad8c587f522e713 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/level/CollisionGetter.java
+++ b/src/main/java/net/minecraft/world/level/CollisionGetter.java
@@ -35,31 +35,33 @@ public interface CollisionGetter extends BlockGetter {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
return this.isUnobstructed(entity, Shapes.create(entity.getBoundingBox()));
}
+ // Paper start - optimise collisions
+ default boolean noCollision(Entity entity, AABB box, boolean loadChunks) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ return !io.papermc.paper.util.CollisionUtil.getCollisionsForBlocksOrWorldBorder(this, entity, box, null, loadChunks, false, entity != null, true, null)
+ && !io.papermc.paper.util.CollisionUtil.getEntityHardCollisions(this, entity, box, null, true, null);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ // Paper end - optimise collisions
+
default boolean noCollision(AABB box) {
- return this.noCollision((Entity)null, box);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // Paper start - optimise collisions
+ return !io.papermc.paper.util.CollisionUtil.getCollisionsForBlocksOrWorldBorder(this, null, box, null, false, false, false, true, null)
+ && !io.papermc.paper.util.CollisionUtil.getEntityHardCollisions(this, null, box, null, true, null);
+ // Paper end - optimise collisions
}
default boolean noCollision(Entity entity) {
- return this.noCollision(entity, entity.getBoundingBox());
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // Paper start - optimise collisions
+ AABB box = entity.getBoundingBox();
+ return !io.papermc.paper.util.CollisionUtil.getCollisionsForBlocksOrWorldBorder(this, entity, box, null, false, false, entity != null, true, null)
+ && !io.papermc.paper.util.CollisionUtil.getEntityHardCollisions(this, entity, box, null, true, null);
+ // Paper end - optimise collisions
}
default boolean noCollision(@Nullable Entity entity, AABB box) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
- try { if (entity != null) entity.collisionLoadChunks = true; // Paper
- for(VoxelShape voxelShape : this.getBlockCollisions(entity, box)) {
- if (!voxelShape.isEmpty()) {
- return false;
- }
- }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
- } finally { if (entity != null) entity.collisionLoadChunks = false; } // Paper
-
- if (!this.getEntityCollisions(entity, box).isEmpty()) {
- return false;
- } else if (entity == null) {
- return true;
- } else {
- VoxelShape voxelShape2 = this.borderCollision(entity, box);
- return voxelShape2 == null || !Shapes.joinIsNotEmpty(voxelShape2, Shapes.create(box), BooleanOp.AND);
- }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // Paper start - optimise collisions
+ return !io.papermc.paper.util.CollisionUtil.getCollisionsForBlocksOrWorldBorder(this, entity, box, null, false, false, entity != null, true, null)
+ && !io.papermc.paper.util.CollisionUtil.getEntityHardCollisions(this, entity, box, null, true, null);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // Paper end - optimise collisions
}
List<VoxelShape> getEntityCollisions(@Nullable Entity entity, AABB box);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
diff --git a/src/main/java/net/minecraft/world/level/EntityGetter.java b/src/main/java/net/minecraft/world/level/EntityGetter.java
2022-12-07 22:35:34 +01:00
index 66a5783e2a83c75ca46d1fd6f97d9de733c01a09..d860ddae508f53d06f74d8ae0efdfc500c1ddf07 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/level/EntityGetter.java
+++ b/src/main/java/net/minecraft/world/level/EntityGetter.java
2022-05-29 21:54:25 +02:00
@@ -49,7 +49,7 @@ public interface EntityGetter {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
return true;
} else {
2022-01-26 04:43:53 +01:00
for(Entity entity : this.getEntities(except, shape.bounds())) {
- if (!entity.isRemoved() && entity.blocksBuilding && (except == null || !entity.isPassengerOfSameVehicle(except)) && Shapes.joinIsNotEmpty(shape, Shapes.create(entity.getBoundingBox()), BooleanOp.AND)) {
+ if (!entity.isRemoved() && entity.blocksBuilding && (except == null || !entity.isPassengerOfSameVehicle(except)) && shape.intersects(entity.getBoundingBox())) { // Paper
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
return false;
}
}
2022-05-29 21:54:25 +02:00
@@ -67,7 +67,7 @@ public interface EntityGetter {
return List.of();
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
} else {
Predicate<Entity> predicate = entity == null ? EntitySelector.CAN_BE_COLLIDED_WITH : EntitySelector.NO_SPECTATORS.and(entity::canCollideWith);
- List<Entity> list = this.getEntities(entity, box.inflate(1.0E-7D), predicate);
+ List<Entity> list = this.getEntities(entity, box.inflate(-1.0E-7D), predicate); // Paper - needs to be negated, or else we get things we don't collide with
if (list.isEmpty()) {
return List.of();
} else {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
diff --git a/src/main/java/net/minecraft/world/level/block/state/BlockBehaviour.java b/src/main/java/net/minecraft/world/level/block/state/BlockBehaviour.java
2022-12-07 22:35:34 +01:00
index efbcd2b78d75481793e0f29a7431d3eb76489abc..9ebc53d434737c8cd39073470b2b5fcbad167812 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/level/block/state/BlockBehaviour.java
+++ b/src/main/java/net/minecraft/world/level/block/state/BlockBehaviour.java
2022-12-07 22:35:34 +01:00
@@ -762,6 +762,12 @@ public abstract class BlockBehaviour implements FeatureElement {
return this.conditionallyFullOpaque;
}
// Paper end - starlight
+ private long blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_SPECIAL_BLOCK;
+
+ public final long getBlockCollisionBehavior() {
+ return this.blockCollisionBehavior;
+ }
+ // Paper end
public void initCache() {
2022-12-07 22:35:34 +01:00
this.fluidState = ((Block) this.owner).getFluidState(this.asState());
@@ -772,6 +778,35 @@ public abstract class BlockBehaviour implements FeatureElement {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
this.shapeExceedsCube = this.cache == null || this.cache.largeCollisionShape; // Paper - moved from actual method to here
Rewrite chunk system (#8177) Patch documentation to come Issues with the old system that are fixed now: - World generation does not scale with cpu cores effectively. - Relies on the main thread for scheduling and maintaining chunk state, dropping chunk load/generate rates at lower tps. - Unreliable prioritisation of chunk gen/load calls that block the main thread. - Shutdown logic is utterly unreliable, as it has to wait for all chunks to unload - is it guaranteed that the chunk system is in a state on shutdown that it can reliably do this? Watchdog shutdown also typically failed due to thread checks, which is now resolved. - Saving of data is not unified (i.e can save chunk data without saving entity data, poses problems for desync if shutdown is really abnormal. - Entities are not loaded with chunks. This caused quite a bit of headache for Chunk#getEntities API, but now the new chunk system loads entities with chunks so that they are ready whenever the chunk loads in. Effectively brings the behavior back to 1.16 era, but still storing entities in their own separate regionfiles. The above list is not complete. The patch documentation will complete it. New chunk system hard relies on starlight and dataconverter, and most importantly the new concurrent utilities in ConcurrentUtil. Some of the old async chunk i/o interface (i.e the old file io thread reroutes _some_ calls to the new file io thread) is kept for plugin compat reasons. It will be removed in the next major version of minecraft. The old legacy chunk system patches have been moved to the removed folder in case we need them again.
2022-09-26 10:02:51 +02:00
this.opacityIfCached = this.cache == null || this.isConditionallyFullOpaque() ? -1 : this.cache.lightBlock; // Paper - starlight - cache opacity for light
+ // Paper start
+ if (io.papermc.paper.util.CollisionUtil.isSpecialCollidingBlock(this)) {
+ this.blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_SPECIAL_BLOCK;
+ } else {
+ try {
+ // There is NOTHING HACKY ABOUT THIS AT ALLLLLLLLLLLLLLL
+ VoxelShape constantShape = this.getCollisionShape(null, null, null);
+ if (constantShape == null) {
+ this.blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_UNKNOWN_BLOCK;
+ } else {
+ constantShape = constantShape.optimize();
+ if (constantShape.isEmpty()) {
+ this.blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_EMPTY_BLOCK;
+ } else {
+ final List<net.minecraft.world.phys.AABB> boxes = constantShape.toAabbs();
+ if (constantShape == net.minecraft.world.phys.shapes.Shapes.getFullUnoptimisedCube() || (boxes.size() == 1 && boxes.get(0).equals(net.minecraft.world.phys.shapes.Shapes.BLOCK_OPTIMISED.aabb))) {
+ this.blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_FULL_BLOCK;
+ } else {
+ this.blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_UNKNOWN_BLOCK;
+ }
+ }
+ }
+ } catch (final Error error) {
+ throw error;
+ } catch (final Throwable throwable) {
+ this.blockCollisionBehavior = io.papermc.paper.util.CollisionUtil.KNOWN_UNKNOWN_BLOCK;
+ }
+ }
+ // Paper end
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
public Block getBlock() {
diff --git a/src/main/java/net/minecraft/world/level/chunk/LevelChunkSection.java b/src/main/java/net/minecraft/world/level/chunk/LevelChunkSection.java
2022-07-21 21:53:04 +02:00
index c3f1334b2bb97f0633f3ea43b97ee49adfd8bc0d..b0c9fce9d4e06cac139e341d218d0b6aac1f1943 100644
--- a/src/main/java/net/minecraft/world/level/chunk/LevelChunkSection.java
+++ b/src/main/java/net/minecraft/world/level/chunk/LevelChunkSection.java
@@ -46,6 +46,110 @@ public class LevelChunkSection {
2022-03-01 06:43:03 +01:00
this.biomes = new PalettedContainer<>(biomeRegistry.asHolderIdMap(), biomeRegistry.getHolderOrThrow(Biomes.PLAINS), PalettedContainer.Strategy.SECTION_BIOMES, null); // Paper - Anti-Xray - Add preset biomes
}
+ // Paper start
+ protected int specialCollidingBlocks;
+ // blockIndex = x | (z << 4) | (y << 8)
+ private long[] knownBlockCollisionData;
+
+ private long[] initKnownDataField() {
+ return this.knownBlockCollisionData = new long[16 * 16 * 16 * 2 / Long.SIZE];
+ }
+
+ public final boolean hasSpecialCollidingBlocks() {
+ return this.specialCollidingBlocks != 0;
+ }
+
+ public static long getKnownBlockInfo(final int blockIndex, final long value) {
+ final int valueShift = (blockIndex & (Long.SIZE / 2 - 1));
+
+ return (value >>> (valueShift << 1)) & 0b11L;
+ }
+
+ public final long getKnownBlockInfo(final int blockIndex) {
+ if (this.knownBlockCollisionData == null) {
+ return 0L;
+ }
+
+ final int arrayIndex = (blockIndex >>> (6 - 1)); // blockIndex / (64/2)
+ final int valueShift = (blockIndex & (Long.SIZE / 2 - 1));
+
+ final long value = this.knownBlockCollisionData[arrayIndex];
+
+ return (value >>> (valueShift << 1)) & 0b11L;
+ }
+
+ // important detail: this returns 32 values, one for localZ = localZ & (~1) and one for localZ = localZ | 1
+ // the even localZ is the lower 32 bits, the odd is the upper 32 bits
+ public final long getKnownBlockInfoHorizontalRaw(final int localY, final int localZ) {
+ if (this.knownBlockCollisionData == null) {
+ return 0L;
+ }
+
+ final int horizontalIndex = (localZ << 4) | (localY << 8);
+ return this.knownBlockCollisionData[horizontalIndex >>> (6 - 1)];
+ }
+
+ private void initBlockCollisionData() {
+ this.specialCollidingBlocks = 0;
+ // In 1.18 all sections will be initialised, whether or not they have blocks (fucking stupid btw)
+ // This means we can't aggressively initialise the backing long[], or else memory usage will just skyrocket.
+ // So only init if we contain non-empty blocks.
+ if (this.nonEmptyBlockCount == 0) {
+ this.knownBlockCollisionData = null;
+ return;
+ }
+ this.initKnownDataField();
+ for (int index = 0; index < (16 * 16 * 16); ++index) {
+ final BlockState state = this.states.get(index);
+ this.setKnownBlockInfo(index, state);
+ if (io.papermc.paper.util.CollisionUtil.isSpecialCollidingBlock(state)) {
+ ++this.specialCollidingBlocks;
+ }
+ }
+ }
+
+ // only use for initBlockCollisionData
+ private void setKnownBlockInfo(final int blockIndex, final BlockState blockState) {
+ final int arrayIndex = (blockIndex >>> (6 - 1)); // blockIndex / (64/2)
+ final int valueShift = (blockIndex & (Long.SIZE / 2 - 1)) << 1;
+
+ long value = this.knownBlockCollisionData[arrayIndex];
+
+ value &= ~(0b11L << valueShift);
+ value |= blockState.getBlockCollisionBehavior() << valueShift;
+
+ this.knownBlockCollisionData[arrayIndex] = value;
+ }
+
+ public void updateKnownBlockInfo(final int blockIndex, final BlockState from, final BlockState to) {
+ if (io.papermc.paper.util.CollisionUtil.isSpecialCollidingBlock(from)) {
+ --this.specialCollidingBlocks;
+ }
+ if (io.papermc.paper.util.CollisionUtil.isSpecialCollidingBlock(to)) {
+ ++this.specialCollidingBlocks;
+ }
+
+ if (this.nonEmptyBlockCount == 0) {
+ this.knownBlockCollisionData = null;
+ return;
+ }
+
+ if (this.knownBlockCollisionData == null) {
+ this.initKnownDataField();
+ }
+
+ final int arrayIndex = (blockIndex >>> (6 - 1)); // blockIndex / (64/2)
+ final int valueShift = (blockIndex & (Long.SIZE / 2 - 1)) << 1;
+
+ long value = this.knownBlockCollisionData[arrayIndex];
+
+ value &= ~(0b11L << valueShift);
+ value |= to.getBlockCollisionBehavior() << valueShift;
+
+ this.knownBlockCollisionData[arrayIndex] = value;
+ }
+ // Paper end
+
public static int getBottomBlockY(int chunkPos) {
return chunkPos << 4;
}
@@ -70,8 +174,8 @@ public class LevelChunkSection {
return this.setBlockState(x, y, z, state, true);
}
- public BlockState setBlockState(int x, int y, int z, BlockState state, boolean lock) {
- BlockState iblockdata1;
+ public BlockState setBlockState(int x, int y, int z, BlockState state, boolean lock) { // Paper - state -> new state
+ BlockState iblockdata1; // Paper - iblockdata1 -> oldState
if (lock) {
iblockdata1 = (BlockState) this.states.getAndSet(x, y, z, state);
@@ -110,6 +214,7 @@ public class LevelChunkSection {
++this.tickingFluidCount;
}
+ this.updateKnownBlockInfo(x | (z << 4) | (y << 8), iblockdata1, state); // Paper
return iblockdata1;
}
@@ -159,6 +264,7 @@ public class LevelChunkSection {
});
2022-03-01 06:43:03 +01:00
// Paper end
+ this.initBlockCollisionData(); // Paper
}
public PalettedContainer<BlockState> getStates() {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
diff --git a/src/main/java/net/minecraft/world/phys/AABB.java b/src/main/java/net/minecraft/world/phys/AABB.java
index 120498a39b7ca7aee9763084507508d4a1c425aa..68cc6f2a78a06293a29317fda72ab3ee79b3533a 100644
--- a/src/main/java/net/minecraft/world/phys/AABB.java
+++ b/src/main/java/net/minecraft/world/phys/AABB.java
@@ -25,6 +25,17 @@ public class AABB {
this.maxZ = Math.max(z1, z2);
}
+ // Paper start
+ public AABB(double minX, double minY, double minZ, double maxX, double maxY, double maxZ, boolean dummy) {
+ this.minX = minX;
+ this.minY = minY;
+ this.minZ = minZ;
+ this.maxX = maxX;
+ this.maxY = maxY;
+ this.maxZ = maxZ;
+ }
+ // Paper end
+
public AABB(BlockPos pos) {
this((double)pos.getX(), (double)pos.getY(), (double)pos.getZ(), (double)(pos.getX() + 1), (double)(pos.getY() + 1), (double)(pos.getZ() + 1));
}
diff --git a/src/main/java/net/minecraft/world/phys/shapes/ArrayVoxelShape.java b/src/main/java/net/minecraft/world/phys/shapes/ArrayVoxelShape.java
2022-05-29 21:54:25 +02:00
index 9d627b8e6bf3140b894d38b9a720896e2d776369..ca5f01be5d5ccfcc56780ff93cca3824409ffc0d 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/phys/shapes/ArrayVoxelShape.java
+++ b/src/main/java/net/minecraft/world/phys/shapes/ArrayVoxelShape.java
@@ -6,6 +6,9 @@ import java.util.Arrays;
import net.minecraft.Util;
import net.minecraft.core.Direction;
+// Paper start
+import it.unimi.dsi.fastutil.doubles.AbstractDoubleList;
+// Paper end
public class ArrayVoxelShape extends VoxelShape {
private final DoubleList xs;
private final DoubleList ys;
@@ -16,6 +19,11 @@ public class ArrayVoxelShape extends VoxelShape {
}
ArrayVoxelShape(DiscreteVoxelShape shape, DoubleList xPoints, DoubleList yPoints, DoubleList zPoints) {
+ // Paper start - optimise multi-aabb shapes
+ this(shape, xPoints, yPoints, zPoints, null, 0.0, 0.0, 0.0);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+ ArrayVoxelShape(DiscreteVoxelShape shape, DoubleList xPoints, DoubleList yPoints, DoubleList zPoints, net.minecraft.world.phys.AABB[] boundingBoxesRepresentation, double offsetX, double offsetY, double offsetZ) {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ // Paper end - optimise multi-aabb shapes
super(shape);
int i = shape.getXSize() + 1;
int j = shape.getYSize() + 1;
@@ -27,6 +35,12 @@ public class ArrayVoxelShape extends VoxelShape {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
} else {
throw (IllegalArgumentException)Util.pauseInIde(new IllegalArgumentException("Lengths of point arrays must be consistent with the size of the VoxelShape."));
}
+ // Paper start - optimise multi-aabb shapes
+ this.boundingBoxesRepresentation = boundingBoxesRepresentation == null ? this.toAabbs().toArray(EMPTY) : boundingBoxesRepresentation;
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ this.offsetX = offsetX;
+ this.offsetY = offsetY;
+ this.offsetZ = offsetZ;
+ // Paper end - optimise multi-aabb shapes
}
@Override
@@ -42,4 +56,152 @@ public class ArrayVoxelShape extends VoxelShape {
2022-05-29 21:54:25 +02:00
throw new IllegalArgumentException();
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
}
+
+ // Paper start
+ public static final class DoubleListOffsetExposed extends AbstractDoubleList {
+
+ public final DoubleArrayList list;
+ public final double offset;
+
+ public DoubleListOffsetExposed(final DoubleArrayList list, final double offset) {
+ this.list = list;
+ this.offset = offset;
+ }
+
+ @Override
+ public double getDouble(final int index) {
+ return this.list.getDouble(index) + this.offset;
+ }
+
+ @Override
+ public int size() {
+ return this.list.size();
+ }
+ }
+
+ static final net.minecraft.world.phys.AABB[] EMPTY = new net.minecraft.world.phys.AABB[0];
+ final net.minecraft.world.phys.AABB[] boundingBoxesRepresentation;
+
+ final double offsetX;
+ final double offsetY;
+ final double offsetZ;
+
+ public final net.minecraft.world.phys.AABB[] getBoundingBoxesRepresentation() {
+ return this.boundingBoxesRepresentation;
+ }
+
+ public final double getOffsetX() {
+ return this.offsetX;
+ }
+
+ public final double getOffsetY() {
+ return this.offsetY;
+ }
+
+ public final double getOffsetZ() {
+ return this.offsetZ;
+ }
+
+ @Override
+ public java.util.List<net.minecraft.world.phys.AABB> toAabbs() {
2021-12-05 10:04:07 +01:00
+ if (this.boundingBoxesRepresentation == null) {
+ return super.toAabbs();
2021-12-05 10:04:07 +01:00
+ }
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ java.util.List<net.minecraft.world.phys.AABB> ret = new java.util.ArrayList<>(this.boundingBoxesRepresentation.length);
+
+ double offX = this.offsetX;
+ double offY = this.offsetY;
+ double offZ = this.offsetZ;
+
+ for (net.minecraft.world.phys.AABB boundingBox : this.boundingBoxesRepresentation) {
+ ret.add(boundingBox.move(offX, offY, offZ));
+ }
+
+ return ret;
+ }
+
+ protected static DoubleArrayList getList(DoubleList from) {
+ if (from instanceof DoubleArrayList) {
+ return (DoubleArrayList)from;
+ } else {
+ return DoubleArrayList.wrap(from.toDoubleArray());
+ }
+ }
+
+ @Override
+ public VoxelShape move(double x, double y, double z) {
+ if (x == 0.0 && y == 0.0 && z == 0.0) {
+ return this;
+ }
+ DoubleListOffsetExposed xPoints, yPoints, zPoints;
+ double offsetX, offsetY, offsetZ;
+
+ if (this.xs instanceof DoubleListOffsetExposed) {
+ xPoints = new DoubleListOffsetExposed(((DoubleListOffsetExposed)this.xs).list, offsetX = this.offsetX + x);
+ yPoints = new DoubleListOffsetExposed(((DoubleListOffsetExposed)this.ys).list, offsetY = this.offsetY + y);
+ zPoints = new DoubleListOffsetExposed(((DoubleListOffsetExposed)this.zs).list, offsetZ = this.offsetZ + z);
+ } else {
+ xPoints = new DoubleListOffsetExposed(getList(this.xs), offsetX = x);
+ yPoints = new DoubleListOffsetExposed(getList(this.ys), offsetY = y);
+ zPoints = new DoubleListOffsetExposed(getList(this.zs), offsetZ = z);
+ }
+
+ return new ArrayVoxelShape(this.shape, xPoints, yPoints, zPoints, this.boundingBoxesRepresentation, offsetX, offsetY, offsetZ);
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
+ }
+
+ @Override
+ public final boolean intersects(net.minecraft.world.phys.AABB axisalingedbb) {
+ // this can be optimised by checking an "overall shape" first, but not needed
+ double offX = this.offsetX;
+ double offY = this.offsetY;
+ double offZ = this.offsetZ;
+
+ for (net.minecraft.world.phys.AABB boundingBox : this.boundingBoxesRepresentation) {
+ if (io.papermc.paper.util.CollisionUtil.voxelShapeIntersect(axisalingedbb, boundingBox.minX + offX, boundingBox.minY + offY, boundingBox.minZ + offZ,
+ boundingBox.maxX + offX, boundingBox.maxY + offY, boundingBox.maxZ + offZ)) {
+ return true;
+ }
+ }
+
+ return false;
+ }
+
+ @Override
+ public void forAllBoxes(Shapes.DoubleLineConsumer doubleLineConsumer) {
+ if (this.boundingBoxesRepresentation == null) {
+ super.forAllBoxes(doubleLineConsumer);
+ return;
+ }
+ for (final net.minecraft.world.phys.AABB boundingBox : this.boundingBoxesRepresentation) {
+ doubleLineConsumer.consume(boundingBox.minX + this.offsetX, boundingBox.minY + this.offsetY, boundingBox.minZ + this.offsetZ,
+ boundingBox.maxX + this.offsetX, boundingBox.maxY + this.offsetY, boundingBox.maxZ + this.offsetZ);
+ }
+ }
+
+ @Override
+ public VoxelShape optimize() {
+ if (this == Shapes.empty() || this.boundingBoxesRepresentation.length == 0) {
+ return this;
+ }
+
+ VoxelShape simplified = Shapes.empty();
+ for (final net.minecraft.world.phys.AABB boundingBox : this.boundingBoxesRepresentation) {
+ simplified = Shapes.joinUnoptimized(simplified, Shapes.box(boundingBox.minX + this.offsetX, boundingBox.minY + this.offsetY, boundingBox.minZ + this.offsetZ,
+ boundingBox.maxX + this.offsetX, boundingBox.maxY + this.offsetY, boundingBox.maxZ + this.offsetZ), BooleanOp.OR);
+ }
+
+ if (!(simplified instanceof ArrayVoxelShape)) {
+ return simplified;
+ }
+
+ final net.minecraft.world.phys.AABB[] boundingBoxesRepresentation = ((ArrayVoxelShape)simplified).getBoundingBoxesRepresentation();
+
+ if (boundingBoxesRepresentation.length == 1) {
+ return new io.papermc.paper.voxel.AABBVoxelShape(boundingBoxesRepresentation[0]).optimize();
+ }
+
+ return simplified;
+ }
+ // Paper end
+
}
diff --git a/src/main/java/net/minecraft/world/phys/shapes/Shapes.java b/src/main/java/net/minecraft/world/phys/shapes/Shapes.java
index 9176735c08a75854209f24113b0e78332249dc4d..731c7dd15f131dc124be6af8f342b122cb89491b 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/phys/shapes/Shapes.java
+++ b/src/main/java/net/minecraft/world/phys/shapes/Shapes.java
@@ -19,16 +19,17 @@ public final class Shapes {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
DiscreteVoxelShape discreteVoxelShape = new BitSetDiscreteVoxelShape(1, 1, 1);
discreteVoxelShape.fill(0, 0, 0);
return new CubeVoxelShape(discreteVoxelShape);
- });
+ }); public static VoxelShape getFullUnoptimisedCube() { return BLOCK; } // Paper - OBFHELPER
public static final VoxelShape INFINITY = box(Double.NEGATIVE_INFINITY, Double.NEGATIVE_INFINITY, Double.NEGATIVE_INFINITY, Double.POSITIVE_INFINITY, Double.POSITIVE_INFINITY, Double.POSITIVE_INFINITY);
private static final VoxelShape EMPTY = new ArrayVoxelShape(new BitSetDiscreteVoxelShape(0, 0, 0), (DoubleList)(new DoubleArrayList(new double[]{0.0D})), (DoubleList)(new DoubleArrayList(new double[]{0.0D})), (DoubleList)(new DoubleArrayList(new double[]{0.0D})));
+ public static final io.papermc.paper.voxel.AABBVoxelShape BLOCK_OPTIMISED = new io.papermc.paper.voxel.AABBVoxelShape(new AABB(0.0, 0.0, 0.0, 1.0, 1.0, 1.0)); // Paper
public static VoxelShape empty() {
return EMPTY;
}
public static VoxelShape block() {
- return BLOCK;
+ return BLOCK_OPTIMISED; // Paper
}
public static VoxelShape box(double minX, double minY, double minZ, double maxX, double maxY, double maxZ) {
@@ -41,29 +42,14 @@ public final class Shapes {
public static VoxelShape create(double minX, double minY, double minZ, double maxX, double maxY, double maxZ) {
if (!(maxX - minX < 1.0E-7D) && !(maxY - minY < 1.0E-7D) && !(maxZ - minZ < 1.0E-7D)) {
- int i = findBits(minX, maxX);
- int j = findBits(minY, maxY);
- int k = findBits(minZ, maxZ);
- if (i >= 0 && j >= 0 && k >= 0) {
- if (i == 0 && j == 0 && k == 0) {
- return block();
- } else {
- int l = 1 << i;
- int m = 1 << j;
- int n = 1 << k;
- BitSetDiscreteVoxelShape bitSetDiscreteVoxelShape = BitSetDiscreteVoxelShape.withFilledBounds(l, m, n, (int)Math.round(minX * (double)l), (int)Math.round(minY * (double)m), (int)Math.round(minZ * (double)n), (int)Math.round(maxX * (double)l), (int)Math.round(maxY * (double)m), (int)Math.round(maxZ * (double)n));
- return new CubeVoxelShape(bitSetDiscreteVoxelShape);
- }
- } else {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
- return new ArrayVoxelShape(BLOCK.shape, (DoubleList)DoubleArrayList.wrap(new double[]{minX, maxX}), (DoubleList)DoubleArrayList.wrap(new double[]{minY, maxY}), (DoubleList)DoubleArrayList.wrap(new double[]{minZ, maxZ}));
- }
+ return new io.papermc.paper.voxel.AABBVoxelShape(new AABB(minX, minY, minZ, maxX, maxY, maxZ)); // Paper
} else {
return empty();
}
}
public static VoxelShape create(AABB box) {
- return create(box.minX, box.minY, box.minZ, box.maxX, box.maxY, box.maxZ);
+ return new io.papermc.paper.voxel.AABBVoxelShape(box); // Paper
}
@VisibleForTesting
@@ -125,6 +111,20 @@ public final class Shapes {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
public static boolean joinIsNotEmpty(VoxelShape shape1, VoxelShape shape2, BooleanOp predicate) {
+ // Paper start - optimise voxelshape
+ if (predicate == BooleanOp.AND) {
+ if (shape1 instanceof io.papermc.paper.voxel.AABBVoxelShape && shape2 instanceof io.papermc.paper.voxel.AABBVoxelShape) {
+ return io.papermc.paper.util.CollisionUtil.voxelShapeIntersect(((io.papermc.paper.voxel.AABBVoxelShape)shape1).aabb, ((io.papermc.paper.voxel.AABBVoxelShape)shape2).aabb);
+ } else if (shape1 instanceof io.papermc.paper.voxel.AABBVoxelShape && shape2 instanceof ArrayVoxelShape) {
+ return ((ArrayVoxelShape)shape2).intersects(((io.papermc.paper.voxel.AABBVoxelShape)shape1).aabb);
+ } else if (shape2 instanceof io.papermc.paper.voxel.AABBVoxelShape && shape1 instanceof ArrayVoxelShape) {
+ return ((ArrayVoxelShape)shape1).intersects(((io.papermc.paper.voxel.AABBVoxelShape)shape2).aabb);
+ }
+ }
+ return joinIsNotEmptyVanilla(shape1, shape2, predicate);
+ }
+ public static boolean joinIsNotEmptyVanilla(VoxelShape shape1, VoxelShape shape2, BooleanOp predicate) {
+ // Paper end - optimise voxelshape
if (predicate.apply(false, false)) {
throw (IllegalArgumentException)Util.pauseInIde(new IllegalArgumentException());
} else {
@@ -196,6 +196,43 @@ public final class Shapes {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
public static VoxelShape getFaceShape(VoxelShape shape, Direction direction) {
+ // Paper start - optimise shape creation here for lighting, as this shape is going to be used
+ // for transparency checks
+ if (shape == BLOCK || shape == BLOCK_OPTIMISED) {
+ return BLOCK_OPTIMISED;
+ } else if (shape == empty()) {
+ return empty();
+ }
+
+ if (shape instanceof io.papermc.paper.voxel.AABBVoxelShape) {
+ final AABB box = ((io.papermc.paper.voxel.AABBVoxelShape)shape).aabb;
+ switch (direction) {
+ case WEST: // -X
+ case EAST: { // +X
+ final boolean useEmpty = direction == Direction.EAST ? !DoubleMath.fuzzyEquals(box.maxX, 1.0, io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON) :
+ !DoubleMath.fuzzyEquals(box.minX, 0.0, io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON);
+ return useEmpty ? empty() : new io.papermc.paper.voxel.AABBVoxelShape(new AABB(0.0, box.minY, box.minZ, 1.0, box.maxY, box.maxZ)).optimize();
+ }
+ case DOWN: // -Y
+ case UP: { // +Y
+ final boolean useEmpty = direction == Direction.UP ? !DoubleMath.fuzzyEquals(box.maxY, 1.0, io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON) :
+ !DoubleMath.fuzzyEquals(box.minY, 0.0, io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON);
+ return useEmpty ? empty() : new io.papermc.paper.voxel.AABBVoxelShape(new AABB(box.minX, 0.0, box.minZ, box.maxX, 1.0, box.maxZ)).optimize();
+ }
+ case NORTH: // -Z
+ case SOUTH: { // +Z
+ final boolean useEmpty = direction == Direction.SOUTH ? !DoubleMath.fuzzyEquals(box.maxZ, 1.0, io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON) :
+ !DoubleMath.fuzzyEquals(box.minZ,0.0, io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON);
+ return useEmpty ? empty() : new io.papermc.paper.voxel.AABBVoxelShape(new AABB(box.minX, box.minY, 0.0, box.maxX, box.maxY, 1.0)).optimize();
+ }
+ }
+ }
+
+ // fall back to vanilla
+ return getFaceShapeVanilla(shape, direction);
+ }
+ public static VoxelShape getFaceShapeVanilla(VoxelShape shape, Direction direction) {
+ // Paper end
if (shape == block()) {
return block();
} else {
@@ -210,7 +247,7 @@ public final class Shapes {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
i = 0;
}
- return (VoxelShape)(!bl ? empty() : new SliceShape(shape, axis, i));
+ return (VoxelShape)(!bl ? empty() : new SliceShape(shape, axis, i).optimize().optimize()); // Paper - first optimize converts to ArrayVoxelShape, second optimize could convert to AABBVoxelShape
}
}
@@ -235,6 +272,53 @@ public final class Shapes {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
public static boolean faceShapeOccludes(VoxelShape one, VoxelShape two) {
+ // Paper start - try to optimise for the case where the shapes do _not_ occlude
+ // which is _most_ of the time in lighting
+ if (one == getFullUnoptimisedCube() || one == BLOCK_OPTIMISED
+ || two == getFullUnoptimisedCube() || two == BLOCK_OPTIMISED) {
+ return true;
+ }
+ boolean v1Empty = one == empty();
+ boolean v2Empty = two == empty();
+ if (v1Empty && v2Empty) {
+ return false;
+ }
+ if ((one instanceof io.papermc.paper.voxel.AABBVoxelShape || v1Empty)
+ && (two instanceof io.papermc.paper.voxel.AABBVoxelShape || v2Empty)) {
+ if (!v1Empty && !v2Empty && (one != two)) {
+ AABB boundingBox1 = ((io.papermc.paper.voxel.AABBVoxelShape)one).aabb;
+ AABB boundingBox2 = ((io.papermc.paper.voxel.AABBVoxelShape)two).aabb;
+ // can call it here in some cases
+
+ // check overall bounding box
+ double minY = Math.min(boundingBox1.minY, boundingBox2.minY);
+ double maxY = Math.max(boundingBox1.maxY, boundingBox2.maxY);
+ if (minY > io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON || maxY < (1 - io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON)) {
+ return false;
+ }
+ double minX = Math.min(boundingBox1.minX, boundingBox2.minX);
+ double maxX = Math.max(boundingBox1.maxX, boundingBox2.maxX);
+ if (minX > io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON || maxX < (1 - io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON)) {
+ return false;
+ }
+ double minZ = Math.min(boundingBox1.minZ, boundingBox2.minZ);
+ double maxZ = Math.max(boundingBox1.maxZ, boundingBox2.maxZ);
+ if (minZ > io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON || maxZ < (1 - io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON)) {
+ return false;
+ }
+ // fall through to full merge check
+ } else {
+ AABB boundingBox = v1Empty ? ((io.papermc.paper.voxel.AABBVoxelShape)two).aabb : ((io.papermc.paper.voxel.AABBVoxelShape)one).aabb;
+ // check if the bounding box encloses the full cube
+ return (boundingBox.minY <= io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON && boundingBox.maxY >= (1 - io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON)) &&
+ (boundingBox.minX <= io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON && boundingBox.maxX >= (1 - io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON)) &&
+ (boundingBox.minZ <= io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON && boundingBox.maxZ >= (1 - io.papermc.paper.util.CollisionUtil.COLLISION_EPSILON));
+ }
+ }
+ return faceShapeOccludesVanilla(one, two);
+ }
+ public static boolean faceShapeOccludesVanilla(VoxelShape one, VoxelShape two) {
+ // Paper end
if (one != block() && two != block()) {
if (one.isEmpty() && two.isEmpty()) {
return false;
diff --git a/src/main/java/net/minecraft/world/phys/shapes/VoxelShape.java b/src/main/java/net/minecraft/world/phys/shapes/VoxelShape.java
index c4ca051720f790f5b8eb860b14e268de8557454d..2182afd1b95acf14c55bddfeec17dae0a63e1f00 100644
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
--- a/src/main/java/net/minecraft/world/phys/shapes/VoxelShape.java
+++ b/src/main/java/net/minecraft/world/phys/shapes/VoxelShape.java
@@ -16,11 +16,17 @@ import net.minecraft.world.phys.BlockHitResult;
import net.minecraft.world.phys.Vec3;
public abstract class VoxelShape {
- protected final DiscreteVoxelShape shape;
+ public final DiscreteVoxelShape shape; // Paper - public
@Nullable
private VoxelShape[] faces;
- VoxelShape(DiscreteVoxelShape voxels) {
+ // Paper start
+ public boolean intersects(AABB shape) {
+ return Shapes.joinIsNotEmpty(this, new io.papermc.paper.voxel.AABBVoxelShape(shape), BooleanOp.AND);
+ }
+ // Paper end
+
+ protected VoxelShape(DiscreteVoxelShape voxels) { // Paper - protected
this.shape = voxels;
}
@@ -163,7 +169,7 @@ public abstract class VoxelShape {
Merge tuinity (#6413) This PR contains all of Tuinity's patches. Very notable ones are: - Highly optimised collisions - Optimised entity lookups by bounding box (Mojang made regressions in 1.17, this brings it back to 1.16) - Starlight https://github.com/PaperMC/Starlight - Rewritten dataconverter system https://github.com/PaperMC/DataConverter - Random block ticking optimisation (wrongly dropped from Paper 1.17) - Chunk ticking optimisations - Anything else I've forgotten in the 60 or so patches If you are a previous Tuinity user, your config will not migrate. You must do it yourself. The config options have simply been moved into paper.yml, so it will be an easy migration. However, please note that the chunk loading options in tuinity.yml are NOT compatible with the options in paper.yml. * Port tuinity, initial patchset * Update gradle to 7.2 jmp said it fixes rebuildpatches not working for me. it fucking better * Completely clean apply * Remove tuinity config, add per player api patch * Remove paper reobf mappings patch * Properly update gradlew * Force clean rebuild * Mark fixups Comments and ATs still need to be done * grep -r "Tuinity" * Fixup * Ensure gameprofile lastaccess is written only under the state lock * update URL for dataconverter * Only clean rebuild tuinity patches might fix merge conflicts * Use UTF-8 for gradlew * Clean rb patches again * Convert block ids used as item ids Neither the converters of pre 1.13 nor DFU handled these cases, as by the time they were written the game at the time didn't consider these ids valid - they would be air. Because of this, some worlds have logspam since only DataConverter (not DFU or legacy converters) will warn when an invalid id has been seen. While quite a few do need to now be considered as air, quite a lot do not. So it makes sense to add conversion for these items, instead of simply suppressing or ignoring the logs. I've now added id -> string conversion for all block ids that could be used as items that existed in the game before 1.7.10 (I have no interest in tracking down the exact version block ids stopped working) that were on https://minecraft-ids.grahamedgecombe.com/ Items that did not directly convert to new items will be instead converted to air: stems, wheat crops, piston head, tripwire wire block * Fix LightPopulated parsing in V1466 The DFU code was checking if the number existed, not if it didn't exist. I misread the original code. * Always parse protochunk light sources unless it is marked as non-lit Chunks not marked as lit will always go through the light engine, so they should always have their block sources parsed. * Update custom names to JSON for players Missed this fix from CB, as it was inside the DataFixers class. I decided to double check all of the CB changes again: DataFixers.java was the only area I missed, as I had inspected all datafixer diffs and implemented them all into DataConverter. I also checked Bootstrap.java again, and re-evaluated their changes. I had previously done this, but determined that they were all bad. The changes to make standing_sign block map to oak_sign block in V1450 is bad, because that's not the item id V1450 accepts. Only in 1.14 did oak_sign even exist, and as expected there is a converter to rename all existing sign items/blocks. The fix to register the portal block under id 1440 is useless, as the flattenning logic will default to the lowest registered id - which is the exact blockstate that CB registers into 1440. So it just doesn't do anything. The extra item ids in the id -> string converter are already added, but I found this from EMC originally. The change for the spawn egg id 23 -> Arrow is just wrong, that id DOES correspond to TippedArrow, NOT Arrow. As expected, the spawn egg already has a dedicated mapping for Arrow, which is id 10 - which was Arrow's entity id. I also ported a fix for the cooked_fished id update. This doesn't really matter since there is already a dataconverter to fix this, but the game didn't accept cooked_fished at the time. So I see no harm. * Review all converters and walkers - Refactor V99 to have helper methods for defining entity/tile entity types - Automatically namespace all ids that should be namespaced. While vanilla never saved non-namespaced data for things that are namespaced, plugins/users might have. - Synchronised the identity ensure map in HelperBlockFlatteningV1450 - Code style consistency - Add missing log warning in V102 for ITEM_NAME type conversion - Use getBoolean instead of getByte - Use ConverterAbstractEntityRename for V143 TippedArrow -> Arrow rename, as it will affect ENTITY_NAME type - Always set isVillager to false in V502 for Zombie - Register V808's converter under subversion 1 like DFU - Register a breakpoint for V1.17.1. In the future, all final versions of major releases will have a breakpoint so that the work required to determine if a converter needs a breakpoint is minimal - Validate that a dataconverter is only registered for a version that is registered - ConverterFlattenTileEntity is actually ConverterFlattenEntity It even registered the converters under TILE_ENTITY, instead of ENTITY. - Fix id comparison in V1492 STRUCTURE_FEATURE renamer - Use ConverterAbstractStatsRename for V1510 stats renamer At the time I had written that class, the abstract renamer didn't exist. - Ensure OwnerUUID is at least set to empty string in V1904 if the ocelot is converted to a cat (this is likely so that it retains a collar) - Use generic read/write for Records in V1946 Records is actually a list, not a map. So reading map was invalid. * Always set light to zero when propagating decrease This fixes an almost infinite loop where light values would be spam queued on a very small subset on blocks. This also likely fixes the memory issues people were seeing. * re-organize patches * Apply and fix conflicts * Revert some patches getChunkAt retains chunks so that plugins don't spam loads revert mc-4 fix will remain unless issues pop up * Shuffle iterated chunks if per player is not enabled Can help with some mob spawning stacking up at locations * Make per player default, migrate all configs * Adjust comments in fixups * Rework config for player chunk loader Old config is not compatible. Move all configs to be under `settings` in paper.yml The player chunk loader has been modified to less aggressively load chunks, but to send chunks at higher rates compared to tuinity. There are new config entries to tune this behavior. * Add back old constructor to CompressionEncoder/Decoder (fixes Tuinity #358) * Raise chunk loading default limits * Reduce worldgen thread workers for lower core count cpus * Raise limits for chunk loading config Also place it under `chunk-loading` * Disable max chunk send rate by default * Fix conflicts and rebuild patches * Drop default send rate again Appears to be still causing problems for no known reason * Raise chunk send limits to 100 per player While a low limit fixes ping issues for some people, most people do not suffer from this issue and thus should not suffer from an extremely slow load-in rate. * Rebase part 1 Autosquash the fixups * Move not implemented up * Fixup mc-dev fixes Missed this one * Rebase per player viewdistance api into the original api patch * Remove old light engine patch part 1 The prioritisation must be kept from it, so that part has been rebased into the priority patch. Part 2 will deal with rebasing all of the patches _after_ * Rebase remaining patches for old light patch removal * Remove other mid tick patch * Remove Optimize-PlayerChunkMap-memory-use-for-visibleChunks.patch Replaced by `Do not copy visible chunks` * Revert AT for Vec3i setX/Y/Z The class is immutable. set should not be exposed * Remove old IntegerUtil class * Replace old CraftChunk#getEntities patch * Remove import for SWMRNibbleArray in ChunkAccess * Finished merge checklist * Remove ensureTickThread impl in urgency patch Co-authored-by: Spottedleaf <Spottedleaf@users.noreply.github.com> Co-authored-by: Jason Penilla <11360596+jpenilla@users.noreply.github.com>
2021-08-31 13:02:11 +02:00
}
}
- private VoxelShape calculateFace(Direction direction) {
+ protected VoxelShape calculateFace(Direction direction) { // Paper
Direction.Axis axis = direction.getAxis();
DoubleList doubleList = this.getCoords(axis);
if (doubleList.size() == 2 && DoubleMath.fuzzyEquals(doubleList.getDouble(0), 0.0D, 1.0E-7D) && DoubleMath.fuzzyEquals(doubleList.getDouble(1), 1.0D, 1.0E-7D)) {