Going forward for new installations, config and data files will be
stored at the platform default paths, as defined by
[env-paths](https://www.npmjs.com/package/env-paths).
For backwards compatibility, if the `~/.waveterm` or `WAVETERM_HOME`
directory exists and contains valid data, it will be used. If this check
fails, then `WAVETERM_DATA_HOME` and `WAVETERM_CONFIG_HOME` will be
used. If these are not defined, then `XDG_DATA_HOME` and
`XDG_CONFIG_HOME` will be used. Finally, if none of these are defined,
the [env-paths](https://www.npmjs.com/package/env-paths) defaults will
be used.
As with the existing app, dev instances will write to `waveterm-dev`
directories, while all others will write to `waveterm`.
Adds new functionality on the backend that will merge any file from the
config directory that matches `<partName>.json` or `<partName>/*.json`
into the corresponding config part (presets, termthemes, etc.). This
lets us separate the AI presets into `presets/ai.json` so that we can
add a dropdown in the AI preset selector that will directly open the
file so a user can edit it more easily. Right now, this will create a
preview block in the layout, but in the future we can look into making
this block disconnected from the layout.
If you put AI presets in the regular presets.json file, it will still
work, since all the presets get merged. Same for any other config part.
Fixes
[https://github.com/wavetermdev/waveterm/security/code-scanning/50](https://github.com/wavetermdev/waveterm/security/code-scanning/50)
To fix the problem, we need to ensure that the size computation for the
allocation does not overflow. This can be achieved by validating the
length of `barr` before performing the arithmetic operation. We will set
a maximum allowable size for `barr` to ensure that the sum of
`oscPrefixLen(oscNum)` and `len(barr)` does not exceed the maximum value
for an `int`.
1. Define a maximum allowable size for `barr` (e.g., 64 MB).
2. Check the length of `barr` against this maximum size before
performing the allocation.
3. If `barr` exceeds the maximum size, return an error.
_Suggested fixes powered by Copilot Autofix. Review carefully before
merging._
---------
Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
If a user has changed the global AI settings and added a different base
url, the Wave Proxy preset would not work because it doesn't unset all
the AI settings.
This provides a checkbox when installing wsh that will prevent the
message from popping up in the future. It can also be disabled by adding
`"askbeforewshinstall": false` to the config file.
Removes global atoms dependency from emain by moving WOS to grab the
globalAtoms from window, if present. Also removes interdependency
between wshrpcutil and wps
Also adds showmenubar setting for Windows and Linux
This will print error messages to the chat when there is an error
getting an ai response. The actual content of the responses are not
forwarded to the models in future requests.
<img width="389" alt="Screenshot 2024-10-09 at 2 36 13 PM"
src="https://github.com/user-attachments/assets/e6c6b1c1-fa19-4456-be3b-596feaeaafed">
This adds a custom resolver to the docsite server to fall back to .html
addresses when it gets a not found for a given path. This fixes an issue
where subpages would return a 404 after a hard reload of the frontend. I
also added an effect that will run on startup to resolve the latest
docsite url and set it in the metadata, since the backend port changes
on every run of the app. I also made it so that setting the default
homepage in the webview widget will also unset any block-specific
homepage.
Adds a meta field `pinnedurl` that can be set to override the
`web:defaulturl` setting for a given block. Also adds a home button to
the webview to reset the block url to the homepage
The help view is now an extension of the webview with some of the chrome
removed.
Also updates the cookie dependency to resolve a vulnerability
This will take the latest artifact from the waveterm-docs repo and embed
it in the app binary. When the help view is launched, it will be served
from our backend. If the embedded copy doesn't exist, such as in
unpackaged versions of the app or in locally packaged versions, it will
use the hosted site instead.
There is a sibling PR in the docs repository to build the embedded
version of the app (strips out some external links, removes Algolia
DocSearch, updates the baseUrl)
https://github.com/wavetermdev/waveterm-docs/pull/46
Only affects windows builds:
Previously, the ~/.waveterm/bin had to exist for wsh to be installed
properly. This fixes it to work on a remote machine where that isn't
already set up.
Fixes issue where closing the last tab in the last window would cause
app to show a blank window the next time it opens. Instead, we should
dispose of the window so that it can be created from scratch the next
time the app launches.
Also fixes a potential DB resource leak from dead windows cluttering up
the DB
json.Unmarshal parses all numbers to float64, which breaks any integer
settings values. This PR changes MetaSettingsType.UnmarshalJSON to use
json.Decoder, which is capable of parsing into a meta-type json.Number,
which can be interpreted as a float or an integer. It also properly
handles pointer types.
## New release flow
1. Run "Bump Version" workflow with the desired version bump and the
prerelease flag set to `true`. This will push a new version bump to the
target branch and create a new git tag.
- See below for more info on how the version bumping works.
2. A new "Build Helper" workflow run will kick off automatically for the
new tag. Once it is complete, test the new build locally by downloading
with the [download
script](https://github.com/wavetermdev/thenextwave/blob/main/scripts/artifacts/download-staged-artifact.sh).
3. Release the new build using the [publish
script](https://github.com/wavetermdev/thenextwave/blob/main/scripts/artifacts/publish-from-staging.sh).
This will trigger electron-updater to distribute the package to beta
users.
4. Run "Bump Version" again with a release bump (either `major`,
`minor`, or `patch`) and the prerelease flag set to `false`.
6. Release the new build to all channels using the [publish
script](https://github.com/wavetermdev/thenextwave/blob/main/scripts/artifacts/publish-from-staging.sh).
This will trigger electron-updater to distribute the package to all
users.
## Change Summary
Creates a new "Bump Version" workflow to manage versioning and tag
creation.
Build Helper is now automated.
### Version bumps
Updates the `version.cjs` script so that an argument can be passed to
trigger a version bump. Under the hood, this utilizes NPM's `semver`
package.
If arguments are present, the version will be bumped.
If only a single argument is given, the following are valid inputs:
- `none`: No-op.
- `patch`: Bumps the patch version.
- `minor`: Bumps the minor version.
- `major`: Bumps the major version.
- '1', 'true': Bumps the prerelease version.
If two arguments are given, the first argument must be either `none`,
`patch`, `minor`, or `major`. The second argument must be `1` or `true`
to bump the prerelease version.
### electron-builder
We are now using the release channels support in electron-builder. This
will automatically detect the channel being built based on the package
version to determine which channel update files need to be generated.
See
[here](https://www.electron.build/tutorials/release-using-channels.html)
for more information.
### Github Actions
#### Bump Version
This adds a new "Bump Version" workflow for managing versioning and
queuing new builds. When run, this workflow will bump the version,
create a new tag, and push the changes to the target branch. There is a
new dropdown when queuing the "Bump Version" workflow to select what
kind of version bump to perform. A bump must always be performed when
running a new build to ensure consistency.
I had to create a GitHub App to grant write permissions to our main
branch for the version bump commits. I've made a separate workflow file
to manage the version bump commits, which should help prevent tampering.
Thanks to using the GitHub API directly, I am able to make these commits
signed!
#### Build Helper
Build Helper is now triggered when new tags are created, rather than
being triggered automatically. This ensures we're always creating
artifacts from known checkpoints.
### Settings
Adds a new `autoupdate:channel` configuration to the settings file. If
unset, the default from the artifact will be used (should correspond to
the channel of the artifact when downloaded).
## Future Work
I want to add a release workflow that will automatically copy over the
corresponding version artifacts to the release bucket when a new GitHub
Release is created.
I also want to separate versions into separate subdirectories in the
release bucket so we can clean them up more-easily.
---------
Co-authored-by: wave-builder <builds@commandline.dev>
Co-authored-by: wave-builder[bot] <181805596+wave-builder[bot]@users.noreply.github.com>
This migrates all remaining eventbus events sent over the websocket to
use the wps interface. WPS is more flexible for registering events and
callbacks and provides support for more reliable unsubscribes and
resubscribes.
Hook into an existing SSH Agent.
This allows us to pull keys already authenticated by the agent and write
to the agent ourselves.
---------
Co-authored-by: Evan Simkowitz <esimkowitz@users.noreply.github.com>
Adds a list of potential remotes to add and filters it as you type. It
also provides options for reconnecting on a disconnection and
specifically connecting to a local connection
The frontend wshserver.ts had a weird circuitous dependency on wos.ts,
which was unnecessary. This moves the misplaced functions into wshrpc.ts
and updates the generation logic.
Adds a new setting for the gap size between tiles in a layout. Also
updates the resize handle calculations so they are dynamically generated
based on the gap size. Also updates the styling for the resize handles
to be more robust.
This also updates the default gap size to 3px.
This also slims out the Block Frame padding so it is just enough that
the blocks don't overlap when there's no gap.
This change shaves ~20 MB off the download size by only copying over the
wavesrv binary that is relevant for whichever architecture we're
currently packaging. This is only relevant for macOS at the moment,
though it can also apply to Windows when we get multi-arch builds
working.
This required renaming our Go binaries from .amd64 to .x64 to comply
with electron-builder's naming conventions.
- Adds connection buttons for previews
- Makes it possible for graphs and previews to connect on backend
(without a terminal open to connection)
- Changes the wsh install message
This keeps the TermSize in RuntimeOpts which allows the terminally to be
correctly sized when reloading a block. Also, it seems to correctly size
the terminal on an app reboot (or at least immediately fixes it).
This fix makes it possible to differentiate between keys when multiple
are provided by the remote server. It does not solve the case of
multiple keys of the same type being shared, but it handles multiple
keys of different types being shared, which is much more common.
Adds a flag to the insert layout action to explicitly set the focus of a
newly inserted node. This also adds a flag in the starter layout to
focus on the terminal block.
This adds a new NodeModel, which can be passed from the TileLayout to
contained blocks. It contains all the layout data that the block should
care about, including focus status, whether a drag operation is
underway, whether the node is magnified, etc.
This also adds a focus stack for the layout, which will let the focus
switch to the last-focused node when the currently-focused one is
closed.
This also addresses a regression in the resize handles that caused them
to be offset from the cursor when dragged.
---------
Co-authored-by: sawka <mike.sawka@gmail.com>
This allows the user to select different connections from the terminal
block. Some features include:
- a status bar at the top of the term block that shows your current
connection
- an icon next to the status bar that shows whether the connection is
currently connected
- the ability to click the status bar and type in a new connection in
order to change the current connection
---------
Co-authored-by: sawka <mike.sawka@gmail.com>
With this PR, Electron will generate a new authorization key that the Go
backend will look for in any incoming requests. The Electron backend
will inject this header with all requests to the backend to ensure no
additional work is required on the frontend.
This also adds a `fetchutil` abstraction that will use the Electron
`net` module when calls are made from the Electron backend to the Go
backend. When using the `node:fetch` module, Electron can't inject
headers to requests. The Electron `net` module is also faster than the
Node module.
This also breaks out platform functions in emain into their own file so
other emain modules can import them.
a couple small bug fixes
- wsh not being executable in windows (this doesn't add it to the path
yet)
- windows using the wrong slash for the path to wsh on the remote
This makes it possible to send wsh commands from wsh on a remote session
to wavesrv running locally. The exact behavior of running those commands
isn't implemented, but the underlying interface is added here.
Adds a new setting that emulates the prefers-reduced-motion media query, allowing users to disable Wave animations without affecting other apps on their system. It also honors the prefers-reduced-motion query in case a system-level configuration is present.
This PR is a large refactoring of the layout code to move as much of the
layout state logic as possible into a unified model class, with atoms
and derived atoms to notify the display logic of changes. It also fixes
some latent bugs in the node resize code, significantly speeds up
response times for resizing and dragging, and sets us up to fully
replace the React-DnD library in the future.
