No entities were lost in this bug, just we were triggering the add entities
before they were loaded due to an inconsistent order of putting chunk into chunkmap.
Any entity that appeared to be gone on the last build will now be back.
This should not ever be used in production!!
This setting is intended for testing so you can try out converting your world
without actually modifying the world files.
This will add some additional overhead to your world, but you're
just testing anyways so that's not a big deal :)
Will store in a folder named after the current version.
PlayerData and Data folders are copied on server start, so there
may be some delay there, but region files are only copied on demand.
This is highly experiemental so backup your world before relying on this to not modify it
1.13 undesirably changed behavior here that chunk load event fired
before the entities were added to the world.
This means any plugin that spawns entities in chunk load event
causes the entities to be registered to the chunk, and then
added to the world twice.
Moves Entity Add to World to be done anytime a chunk is
registered to the Chunk Map, and ignore other calls.
Fixes#1288
* master:
Always process chunk registration after moving
Always move Entity to its new Chunk even if unloaded
If Entity is added to chunk, look up the chunk if current isnt set
Ignore Dead Entities in entityList iteration
Always process chunk removal in removeEntity
Spigot 1.13 checks if any field (which are manually copied from the ItemStack's "tag" NBT tag) on the ItemMeta class of an ItemStack is set.
We could just check if the "tag" NBT tag is empty, albeit that would break some plugins. The only general tag added on 1.13 is "Damage", and we can just check if the "tag" NBT tag contains any other tag that's not "Damage" (https://minecraft.gamepedia.com/Player.dat_format#Item_structure) making the `hasItemStack` method behave as before.
Check the `ItemMetaTest#testTaggedButNotMeta` method to see how this method behaves. (I also added some extra tests).
`hasItemMeta()` will return true if `ItemStack.getDamage() != 0` or it has the `Damage` tag or any other tag is set.
Closes#1222
This will help guarantee that entities are always in the
chunk that they are currently located at.
Should hopefully also fix Citizens triggering the "Saved to wrong chunk" message
Vanilla logic here would allow us to remvoe an entity from
its current chunk, and if it was going to move into an unloaded
chunk, that entity would not be added to the unloaded chunk.
This is bad because this will result in the entity being lost!
In almost all cases, the chunk will be loaded, but in the event
it wasn't, instead of losing the entity, load the chunk to add
the entity to it.
Hopefully will (f)ix #1280...
I'm suspicious that Citizens isn't calling things in the same order and causes the current
chunk to not be set, which then bugs removals. Though this doesn't make any sense to me,
so this likely won't fix it...
But if the isAddedToChunk is true, we really should be returning a chunk anyways if its loaded.
A spigot change delays removal of entities from the entity list.
This causes a change in behavior from Vanilla where getEntities type
methods will return dead entities that they shouldn't otherwise be doing.
This will ensure that dead entities are skipped from iteration since
they shouldn't of been in the list in the first place.
Spigot might skip chunk registration changes in removeEntity
which can keep them in the chunk when they shouldnt be if done
during entity ticking.
Should fix some cases where "Entity is still in another chunk section"
Related to #1223
Vanilla logic checks unload queue and overwrites if its in it.
we're triggering this if a chunk unloads, and reloads immediately in same tick.
Added check for unload queue to not treat as duplicate
Also fixed the config setting not even loading
Should fix#1280
Citizens hijacks entity map, and im guessing under the right conditions
the result might actually be null during entity creation
Pre the cache patch, the id is looked up on save, so it was fine.
Now, if its null and the save ID is requested, we will try to look
it up again and cache it if found.
While upstream has now made this event cancellable, their changes
result in the vechicle being removed before the event is called,
thus leading cancellation to not behave as expected.
See https://github.com/PaperMC/Paper/issues/1223
Should fix Vanilla bugs
Minecraft is saving invalid entities to the chunk files.
Avoid saving bad data, and also make improvements to handle
loading these chunks. Any invalid entity will be instant killed,
so lets avoid adding it to the world...
This lets us be safer about the dupe UUID resolver too, as now
we can ignore instant killed entities and avoid risk of duplicating
an invalid entity.
This should reduce log occurrences of dupe uuid messages.
Also reduce the logging spam overall.
Setting the flag updates the spawner's delay which stops the spawner from trying to find a new spawn position each tick efter the event was cancelled/aborted which makes it usable for mob stackers/mergers and other plugins that don't actually want any mob to spawn in the spawner cycle but keep the overall behaviour close to vanilla.
This might slightly effect existing plugins that use this event but I doubt anyone really relied on this behaviour, the only possible use case that I can think of is cancelling the event until you find a suitable position in your plugin... and this should be handled by the plugin itself by cancelling and spawning at the position manually.
CraftBukkit added synchronization to read and write methods. This adds
much more contention on this object for accessing region files, as
the entire read and write of NBT data is now a blocking operation.
This causes issues when something then simply needs to check if a chunk exists
on the main thread, causing a block...
However, this synchronization was unnecessary, because there is already
enough synchronization done to keep things safe
1) Obtaining a Region File: Those methods are still static synchronized.
Meaning we can safely obtain a Region File concurrently.
2) RegionFile data access: Methods reading and manipulating data from
a region file are also marked synchronized, ensuring that no 2 processes
are reading or writing data at the same time.
3) Checking a region file for chunkExists: getOffset is also synchronized
ensuring that even if a chunk is currently being written, it will be safe.
By removing these synchronizations, we reduce the locking to only
when data is being write or read.
GZIP compression and NBT Buffer creation will no longer be part of the
synchronized context, reducing lock times.
Ultimately: This brings us back to Vanilla, which has had no indication of region file loss.
Closes#1260
* master:
Add some debug for entity slices
Mark chunk dirty on entity changes
Reduce and improve dupe uuid resolve message
Add more entity debug info
Bring some 1.13 authors to master
Fixed more stuff
Remove unsed method
Extend player profile API to support skin changes
Extend player profile API to support skin changes
Cleaned up some implementation notes to use existing Vanilla method for some things.
merged into parent patch
7dd5837d Fixed more stuff (NickAcPT)
09f01353 Remove unsed method (NickAcPT)
e5ea4656 Extend player profile API to support skin changes (NickAcPT)
e67d55d0 Extend player profile API to support skin changes (NickAcPT)
* pull/1250/head:
Fixed more stuff
Remove unsed method
Extend player profile API to support skin changes
Extend player profile API to support skin changes
0069113b Put the decompile fixes into MC Dev Fixes patch (Andrew Steinborn)
608b5e52 Optimize RegistryID.c() (Andrew Steinborn)
* pull/1257/head:
Put the decompile fixes into MC Dev Fixes patch
Optimize RegistryID.c()
It's possible we won't hit this on the servers current state since nothing is async,
but we are working towards that.
I experienced a crash due to this code during my work.
Our changes for the spawn radius have the potential to throw an ArithmeticException
should the server be stopped before we've loaded worlds, we check if the server is
running earlier to check if we should even consider attempting to load chunks, which
would cause us to, 1) not load chunks anyways, as we're disabled; 2) throw an
ArithmeticException due to us expecting that we're going to be loading more than 0 chunks.
These chunks are unfinished, and waste cpu time saving these unfinished chunks.
the loadChunk method refuses to acknoledge they exists, and will restart
a new chunk generation process to begin with, so saving them serves no benefit.
* master:
Duplicate UUID Resolve Option
Add more information to Entity.toString
change LAST_EDIT to PAPER_LAST_EDIT for edit commands
Add more information to Entity.toString()
Add Debug Entities option to debug dupe uuid issues
Guard the Entity.SHARED_RANDOM from seed changes
Create a symlink on not-windows to current minecraft decompile dir
Due to a bug in 2e29af3df0
which was added all the way back in March of 2016, it was unknown (potentially not at the time)
that an entity might actually change the seed of the random object.
At some point, EntitySquid did start setting the seed. Due to this shared random, this caused
every entity to use a Random object with a predictable seed.
This has caused entities to potentially generate with the same UUID....
Over the years, servers have had entities disappear, but no sign of trouble
because CraftBukkit removed the log lines indicating that something was wrong.
We have fixed the root issue causing duplicate UUID's, however we now have chunk
files full of entities that have the same UUID as another entity!
When these chunks load, the 2nd entity will not be added to the world correctly.
If that chunk loads in a different order in the future, then it will reverse and the
missing one is now the one added to the world and not the other. This results in very
inconsistent entity behavior.
This change allows you to recover any duplicate entity by generating a new UUID for it.
This also lets you delete them instead if you don't want to risk having new entities added to
the world that you previously did not see.
But for those who are ok with leaving this inconsistent behavior, you may use WARN or NOTHING options.
It is recommended you regenerate the entities, as these were legit entities, and deserve your love.
Added code that refreshes the player's skin by sending packets with a special order, telling the client to respawn the player and re-apply the game profile
Added code that refreshes the player's skin by sending packets with a special order, telling the client to respawn the player and re-apply the game profile
The removal of `ServerConnection.this.h.add(networkmanager);` got
lost in the 1.13 update, causing network managers to be registered
twice.
Fixes "handleDisconnection() called twice" warning spam in console.
Some of the fields in the anonymous class are named the same as the
surrounding method's parameters, which caused the fields to be
initialized incorrectly.
That way it keeps returning the same block position, resulting
in an infinite loop during chunk generation.
* master:
Don't process despawn if entity is in a chunk scheduled for unload
Fix Squids corrupting the entire servers entity randomness....
Fix placement of chunk tracking - Fixes#1199
This won't happen anyways if the user has
"skip ticking for entities in chunks scheduled for unload" turned on,
but if they don't, protect from this instant killing the entity to
keep it vanilla in behavior
a player may teleport away, and trigger instant despawn
Spigot had code that returned early in chunk add/remove methods.
This was causing our code added to set current chunks and counts to
be skipped over if the entity was default not persistent but made persistent.
This was the source of many issues
Fixes#1208
When interacting with entities with an item, the client will assume
the interaction is successful, and update the held item on the
client. However, if the interaction is cancelled on the server side,
the client will still mistakenly remove/replace the item in hand.
Examples for this are milking cows with a bucket or dyeing sheep.
The bucket is replaced with milk and the dye removed from inventory.
Refresh the player inventory when PlayerInteractEntityEvent is
cancelled to avoid this problem.
The adjacent blocks of doors, double plants, pistons and beds need
to be updated manually from the server when cancelling a block break
from a player, as it otherwise causes the other parts to disappear
on the client.
This is already done for doors but only for the BlockBreakEvent,
not for PlayerInteractEvent. Move the code to a common method
and also handle the other blocks in similar ways.
The extra buffer used to decode the strings sent by the client
in the legacy ping protocol was never released. However, creating
an extra copy of the buffer just to decode it to a string isn't
actually necessary: We can just call toString() directly on the
original buffer.
Additionally, free the buffer in handlerRemoved() to handle cases
where the client never sends enough bytes to form a valid legacy
ping request.
Closes#1197
While this really undoes a lot of the desired performance gains avoiding chunk lookups,
we sadly have to accept this because we are seeing lots of bugs with entities.
Allows you to increase how far to check for a safe place to respawn
a player near their bed, allowing a better chance to respawn the
player at their bed should it of became obstructed.
Defaults to vanilla 1.
In many places where we simply want the current chunk the entity
is in, instead of doing a hashmap lookup for it, we now have access
to the object directly on the Entity/TileEntity object we can directly grab.
Use that local value instead to reduce lookups in many hot places.
This enables us a fast reference to the entities current chunk instead
of having to look it up by hashmap lookups.
We also store counts by type to further enable other performance optimizations in later patches.
This is the best way to get an entity when the world and its UUID are known.
It is faster than Server.getEntity(UUID) because it does not have to iterate all worlds
This fixes a CRITICAL missing part of the Bukkit API due to mistakes on upstream
refusing to implement the Sentient NPC baseclass of all NPC's.
Until now, the Bukkit API has not provided a way for accessing and setting
a non creature entities target.
Although Flying, Slime, Ambient, and Water mobs all supported targets internally,
you were unable to get/set it.
Now with the SentientNPC API and these API's moved down, every sentient NPC has
access to target data.
This only impacted people who used our useSnapshots new API in a plugin,
which obviously was no one as the data result was completely broken.
Merged the NPE check patch into mine since it has to handle it too.
This event is called when an entity receives knockback by another entity. The knockback can be modified in the event. If the event is cancelled the entity is not knocked back.
Also renames patch file to better express what it's doing.
It is presumed that those using this config option intend for
suffocation checks to be disabled in all instances. In doing so, they
inherently assume the advantages and issues associated with removing
said safety check.
If the community expresses a desire for more specific options regarding
the handling of this safety feature, we can investigate providing them.
Fixes GH-1149
Called when a player is firing a bow and the server is choosing an arrow to use.
Plugins can skip selection of certain arrows and control which is used.
Rewrites the Vanilla luck application formula so that luck can be
applied to items that do not have any quality defined.
See: https://luckformula.emc.gs for data and details
-----------
The rough summary is:
My goal was that in a pool, when luck was applied, the pool
rebalances so the percentages for bigger items is
lowered and smaller items is boosted.
Do this by boosting and then reducing the weight value,
so that larger numbers are penalized more than smaller numbers.
resulting in a larger reduction of entries for more common
items than the reduction on small weights,
giving smaller weights more of a chance
-----------
This work kind of obsoletes quality, but quality would be useful
for 2 items with same weight that you want luck to impact
in varying directions.
Fishing still falls into that as the weights are closer, so luck
will invalidate junk more.
This change will result in some major changes to fishing formulas.
-----------
I would love to see this change in Vanilla, so Mojang please pull :)
hashCodes are not allowed to change, however bukkit used a value
that does change, the entityId.
When an entity is teleported dimensions, the entity reference is
replaced with a new one with a new entity ID.
For hashCode, we can simply use the UUID's hashCode to keep
the hashCode from changing.
equals() is ok to use getEntityId() because equals() should only
be true if both the left and right are the same reference.
Since entity ids can not duplicate during runtime, this
check is essentially the same as this.getHandle() == other.getHandle()
However, replaced it too to make it clearer of intent.
To teleport an entity between dimensions, the server makes a copy
and puts the copy in the new location, and marks the old one dead.
If this method got called for the same world in the same tick,
the entity would not have been removed from the UUID map, and the
world readd would fail.
This can be triggered even with a plugin if the entity is teleported
twice in the same tick, from world A to B, then back from B to A.
The re-add to A will fail to add the entity to the world. It will
actually be there, but it will not be visible on the client until
the server is restarted to re-try the add to world process again.
This bug was unlikely to be seen by many due to the double teleport
requirement, but plugins (such as my own) use this method to
trigger a "reload" of the entity on the client.
This improves plugins like Citizens that rely on direct instance of Yggdrasil implementations.
Instead of wrapping, directly extend and override the methods.
Went ahead and wrapped all of the services in prep in the base patch, then features modify what they need
Adds ability to control who receives it and who is the source/sender (vanish API)
the standard API is to send the packet to everyone in the world, which is ineffecient.
This adds a new Builder API which is much friendlier to use.
Players are able to use alt accounts and enderpearls to travel
long distances utilizing the pearls in unloaded chunks and loading
the chunk later when convenient.
This disables that by not saving the thrower when the chunk is unloaded.
This is mainly useful for survival servers that do not allow freeform teleporting.
Fires an event anytime an enderman intends to teleport away from the player
You may cancel this, enabling ranged attacks to damage the enderman for example.
Resolves#1101
reporter of this issue was incorrect and did not verify vanilla logic
vanilla logic only skips ticks if the flag is set
spigots change causes bugs as it now skips ticking and processing
chunk teleportation, which was a bug I fixed many many years ago...
Prior to this change the server would crash when attempting to load a
chunk from a region with bad data.
After this change the server will defer back to vanilla behavior. At
this time, that means attempting to generate a chunk in its place
(and occasionally just not generating anything and leaving small
holes in the world).
Should Mojang choose to alter this behavior in the future, this change
will simply defer to whatever that new behavior is.
It is often difficult to diagnose new issues server admins get when
upgrading to a new server version because the only information they are
able to tell us regarding the server version they are running is
"latest". This commit attempts to mitigate this by keeping track of the
previous version of Paper they were running, which is then reported by
the `/version` or `/paper version` command. This gives us a better idea
of the commits included in the upgrade, which may help diagnose new
issues easier.
This change by spigot ensures that many interactins with chunks,
e.g. getting a list of TEs will cause the chunk to be marked for not
unloading and will block their unload. This is especially true for
servers using Timings (it needs to access the TE list of chunks), or
any plugins which need to access entity/TE lists periodically.
At the time this was re-added, there was concern around how the JIT
would handle the system property that enabled it.
This shouldn't be a problem, and as such we no longer need to block
access to it.
The Vanilla Method Profiler will not provide much to most users however
there is no harm in providing it as an option. For most users, the
recommended and supported method for determining performance issues with
Paper will continue to be Timings.
In some enviroments, the channel limit set by spigot can cause issues,
e.g. servers which allow and support the usage of mod packs.
provide an optional flag to disable this check, at your own risk.
* Make the legacy ping handler more reliable
The Minecraft server often fails to respond to old ("legacy") pings
from old Minecraft versions using the protocol used before the switch
to Netty in Minecraft 1.7.
Due to packet fragmentation[1], we might not have all needed bytes
available when the LegacyPingHandler is called. In this case, it will
run into an error, remove the handler and continue using the modern
protocol.
This is unlikely to happen for the first two revisions of the legacy
ping protocol (used in Minecraft 1.5.x and older) since the request
consists of only one or two bytes, but happens frequently for the
last/third revision introduced in Minecraft 1.6.
It has much larger, variable packet sizes due to the inclusion of
the virtual host (the hostname/port used to connect to the server).
The solution[2] is simple: If we find more than two matching bytes,
we buffer the remaining bytes until we have enough to fully read and
respond to the request.
[1]: https://netty.io/wiki/user-guide-for-4.x.html#wiki-h3-11
[2]: https://netty.io/wiki/user-guide-for-4.x.html#wiki-h4-13
* Add legacy ping support to PaperServerListPingEvent
Add a new method to StatusClient check if the client is a legacy
client that does not support all of the features provided in the
event.
Don't want to risk mutating players properties in server list (unlikely, but lets be proper)
and Skull also has a setter API, so that should be used too.
* Drop original implementation for old player sample API
* Add extended PaperServerListPingEvent
Add a new event that extends the original ServerListPingEvent
and allows full control of the response sent to the client.
* Implement deprecated player sample API
I mistakenly thought .complete() also checked for textures, which was not the case
So the logic was not working as desired.
Also some undesired logic paths lead to textures of the logging in player being dropped, forcing
us to always load the textures immediately again on login, leading to rate limits.
Everythings now good
the .complete() api now will default specify to also complete textures, but you may
pass false to it to skip loading textures.
Gets the unique ID of the player currently known as the specified player name
In Offline Mode, will return an Offline UUID
This is a more performant way to obtain a UUID for a name than loading an OfflinePlayer
This ensures we look up the name for ID only Profiles
If the profile is in the UserCache, we can get those details quickly
This should avoid some unnecessary round trips.
Additionally, handle profiles for offline mode to use offline UUID's
Bukkit restricts command execution of signs to test if the sender
has permission to run the specified command. This breaks vanilla
maps that use signs to intentionally run as elevated permission.
Bukkit provides an unrestricted advancements setting, so this setting
compliments that one and allows for unrestricted signs.
We still filter sign update packets to strip out commands at edit phase,
however there is no sanity in ever expecting creative mode to not be
able to create signs with any command.
Creative servers should absolutely never enable this.
Non creative servers, enable at own risk!!!
The Craft Scheduler still uses the primary thread for task scheduling.
This results in the main thread still having to do work as part of the
dispatching of async tasks.
If plugins make use of lots of async tasks, such as particle emitters
that want to keep the logic off the main thread, the main thread still
receives quite a bit of load from processing all of these queued tasks.
Additionally, resizing and managing the pending entries for all of
these asynchronous tasks takes up time on the main thread too.
This commit replaces the implementation of the scheduler when working
with asynchronous tasks, by forwarding calls to the new scheduler.
The Async Scheduler uses a single thread executor for "management" tasks.
The Management Thread is responsible for all adding and dispatching of
scheduled tasks.
The mainThreadHeartbeat will send a heartbeat task to the management thread
with the currentTick value, so that it can find which tasks to execute.
Scheduling of an async tasks also dispatches a management task, ensuring
that any Queue resizing operation occurs off of the main thread.
The async queue uses a complete separate PriorityQueue, ensuring that resize
operations are decoupled from the sync tasks queue.
Additionally, an optimization was made that if a plugin schedules
a single, non repeating, no delay task, that we immediately dispatch it
to the executor pool instead of scheduling it. This avoids an unnecessary
round trip through the queue, as well as will reduce the size growth of the
queue if a plugin schedules lots of asynchronous tasks.
This seems completely pointless, as packet dispatch uses .writeAndFlush.
Things seem to work fine without implicit flushing, but incase issues arise,
provide a System property to re-enable it using improved logic of doing the
flushing on the netty event loop, so it won't do the flush on the main thread.
Renable flushing by passing -Dpaper.implicit-flush=true
This will force the saves to spread over multiple ticks even when many
players auto save interval is aligned, avoiding spikes on large servers.
Closes#1021
Plugins were abusing this to dispatch commands async anyways.
We will no longer check that flag, and force all commands to be ran sync.
Use a different boolean for allowing things go to through on shutdown/restart instead.
Resolves#1004Resolves#1005
- Lots of itemstack cloning removed. Only clone if the item is actually moved
- Return true when a plugin cancels inventory move item event instead of false, as false causes pulls to cycle through all items.
However, pushes do not exhibit the same behavior, so this is not something plugins could of been relying on.
- Add option (Default on) to cooldown hoppers when they fail to move an item due to full inventory
- Skip subsequent InventoryMoveItemEvents if a plugin does not use the item after first event fire for an iteration
This is adds basic item meta for armor stands. It does not add all
possible metadata however.
There are armor, hand, and equipment types, as well as position data
that can also be added here. This initial implementation should serve as
a starting point for future additions in this area.
Fixes GH-559
This is a source of MAJOR lag for hoppers, as well as a gameplay bug.
This removes the necessity to disable the cat on chest behavior to improve performance.
now performance will be improved even if you have cat chest detection on.
Allows plugins to populate profile properties from local sources to avoid calls out to Mojang API
to fill in textures for example.
If Mojang API does need to be hit, event fire so you can get the results.
This is useful for implementing a ProfileCache for Player Skulls
This simply provides the base API to create the objects. Further commits will come that adds
adds usage of this API to existing GameProfile based API's, as well as new API's.
This event can be used for when you want to exclude a certain player
from triggering monster spawns on a server.
Also a highly more effecient way to blanket block spawns in a world
Adds an event to fire before an Entity is created, so that plugins that need to cancel
CreatureSpawnEvent can do so from this event instead.
Cancelling CreatureSpawnEvent rapidly causes a lot of garbage collection and CPU waste
as it's done after the Entity object has been fully created.
Mob Limiting plugins and blanket "ban this type of monster" plugins should use this event
instead and save a lot of server resources.
See: https://github.com/PaperMC/Paper/issues/917
Limit the number of generations that can occur in a single tick, forcing them
to be spread out more.
Defaulting to 10 as an average generation is going to be 3-6ms, which means 10 will
likely cause the server to lose TPS, but constrain how much.
This should result in no noticeable speed reduction in generation for servers not
lagging, and let larger servers reduce this value according to their own desires.
add a system property to allow people to tweak how long the server
will wait for a reply. There is a compromise here between lower and higher
values, lower values will mean that dead connections can be closed sooner,
whereas higher values will make this less sensitive to issues such as spikes
from networking or during connections flood of chunk packets on slower clients,
at the cost of dead connections being kept open for longer.
Instead of overriding add within the queue, never add runnables to the
queue if the light queue is disabled.
This change is made to make timings reports and stacktraces less
confusing for administrators, who prior to this change, would have seen
the lighting queue referenced in both, regardless of whether or not it
was enabled.
This change should not affect performance, nor is it made with the
intent to.
This allows plugins that give players the ability to apply the experience
points to the Item Mending formula, which will repair an item instead
of giving the player experience points.
Both an API To standalone mend, and apply mending logic to .giveExp has been added.
