Updates all collab sounds, add screen sharing sounds.
Release Notes:
- Improved collaboration sounds for joining and leaving a call, muting
and unmuting the mic.
- Added a sound when you start and stop screen sharing.
Current logic does not need to access inlays by id in O(1), future
dynamic hints would need to know which hint they hover at, but that will
be done using binary search over the position's anchor we hover on;
nothing else seems to need this HashMap in the near future.
Because of that removal, no need to store `InlayId` apart from the
`Inlay`, hence remove the `InlayProperties` struct entirely.
This allows to eliminate a few generics along the way.
Release Notes:
- N/A
Just some theme tidying, renames some things to be more consistent with
our planned naming conventions going forward.
Release Notes:
- N/A (No public facing changes)
We removed the `theme_testbench` crate a while back - It seems like that
was the only thing using the `color_scheme` field in the exported theme.
Removing this from the theme removes something like 42k lines of
generated JSON every time we build the theme (2k lines / 28% of the
total lines per generated theme!)
Release Notes:
- N/A (No public facing changes)
This PR adds a theme store to allow components to directly access the
theme without requiring it to be passed down as props every time it is
used.
So before, you might need to do something like `text(theme, "variant",
"hovered")`, you could now just call `text("variant", "hovered")`.
This also means that style_trees don't need to be called with a theme
either:
```ts
export default function app(): any {
const theme = useTheme()
return {
meta: {
name: theme.name,
is_light: theme.is_light,
},
command_palette: command_palette(),
contact_notification: contact_notification(),
// etc...
}
}
```
We do this by creating a zustand store to store the theme, and allow it
to be accessed with `useThemeStore.getState().theme`.
```ts
import { create } from "zustand"
import { ColorScheme } from "./color_scheme"
type ThemeState = {
theme: ColorScheme | undefined
setTheme: (theme: ColorScheme) => void
}
export const useThemeStore = create<ThemeState>((set) => ({
theme: undefined,
setTheme: (theme) => set(() => ({ theme })),
}))
export const useTheme = (): ColorScheme => {
const { theme } = useThemeStore.getState()
if (!theme) throw new Error("Tried to use theme before it was loaded")
return theme
}
```
Release Notes:
- N/A (No public facing changes)
Closes
https://linear.app/zed-industries/issue/Z-2513/panic-in-refresh-inlay-hints
* Filter out queries for outdated buffers just before hint tasks spawn:
multicaret edits might emit standalone events simultaneously
* Only spawn inlay update tasks for visible buffers with corresponding
language
* Do not spawn tasks for local projects' buffers without LSP servers
Release Notes:
- N/A
* Filter out queries for outdated buffers just before hint tasks spawn:
multicared edits might empit standalone events simultaneously
* Only spawn inlay update tasks for visible buffers with corresponding
language
* Do not spawn tasks for local projects' buffers without LSP servers
As part of an optimization in
https://github.com/zed-industries/zed/pull/2663, I changed the way that
the worktree ignores FS events within unloaded directories. But this
accidentally prevented us from detecting some events that occur inside
of `.git` directories.
In this PR, I've made further tweaks to which FS events we can ignore.
We now explicitly opt *in* to scanning `.git` (shallowly) directories
(even though they are ignored). Note that we still don't recursively
scan the git directory (including all of the files inside `objects`
etc). This seems like the correct amount of work to do, and from my
testing (and our unit tests that use the real FS and real git
repositories), it seems to work correctly.
Release Notes:
- Fixed a bug where Zed would not detect some git repository changes
(preview only).
Fixes: https://github.com/zed-industries/community/issues/1712
The keymaps were adding in a `"cmd-enter": "editor::NewlineBelow",`
entry in the context of `Editor`, and this was clobbering the assist
command in the assistant panel context. Zed now defines this command in
the default keymap under the context of `"context": "Editor && mode ==
full"`. All I needed to basically do was remove that command from the
keymaps. I also removed the `"cmd-shift-enter": "editor::NewlineAbove"
from the `Editor` context in those keymaps as wel, as it is also defined
in the default keymap.
Release Notes:
- Fix bug preventing the `assistant: assist` command from working in
certain keymaps
This PR adds a new way to make files / directories in the project panel,
by writing a path instead of a file.
TODO:
- [x] Solve a race condition that sometimes causes the newly created
file to not be selected / expanded correctly.
- [x] Change file refreshes to be minimal
Release Notes:
- Adds the ability to create new folders in the create-file action
([743](https://github.com/zed-industries/community/issues/743))
This PR adds a new mouse event type for catching when a click happens
outside of a given region.
This was added because I noticed a 'race condition' between the context
menu and the buttons which deploy a context menu. Buttons use on
an`on_click()` handler to deploy the context menu, but the context menu
was closing itself with an `on_down_out()` handler. This meant that the
order of operations was:
0. Context menu is open
1. User presses down on the button, _outside of the context menu_
2. `on_down_out()` is fired, closing the context menu
3. User releases the mouse
4. `click()` is fired, checks the state of the context menu, finds that
it's closed, and so opens it
You can see this behavior demonstrated with this video with a long-click
here:
https://github.com/zed-industries/zed/assets/2280405/588234c3-1567-477f-9a12-9e6a70643527
~~Switching from `on_down_out()` to `on_click_out()` means that the
click handler for the button can close the menu before the context menu
gets a chance to close itself.~~
~~However, GPUI does not have an `on_click_out()` event, hence this
PR.~~
~~Here's an example of the new behavior, with the same long-click
action:~~
https://github.com/zed-industries/zed/assets/2280405/a59f4d6f-db24-403f-a281-2c1148499413
Unfortunately, this `click_out` is the incorrect event for this to
happen on. This PR now adds a mechanism for delaying the firing of a
cancel action so that toggle buttons can signal that this on_down event
should not result in a menu closure.
Release Notes:
* Made context menus deployed from buttons toggle, instead of
hide-and-re-show, visibility on click
Part of https://github.com/zed-industries/community/issues/138
Part of https://linear.app/zed-industries/issue/Z-477/inlay-hints
Supports LSP requests for inlay hints, LSP /refresh request to reload
them.
Reworks DisplayMap and underlying layer to unite suggestions with inlay
hints into new, `InlayMap`.
Adds a hint cache inside `Editor` that tracks buffer/project/LSP request
events, updates the hints and ensures opened editors are showing up to
date text hints on top.
Things left to do after this PR:
* docs on how to configure inlay hints
* blogpost
* dynamic hints: resolve, hover, navigation on click, etc.
Release Notes:
- Added basic support of inlay hints
This PR moves the theme / `/styles` typescript app to use snake_case to
better align with the rust app and make it easier to reference things
across both apps.
It also configures ESLint in the styles app and fixes many ESLint
errors.
Going forward from this PR we will use `snake_case` throughout the
theme.
Release Notes:
- N/A (No public facing changes)
Closes
https://linear.app/zed-industries/issue/Z-352/high-memory-usage-from-fs-scanning-if-project-contains-symlinks-that
### Background
Currently, when you open a project, Zed eagerly scans the directory,
building an in-memory representation of all of the files and directories
within. This scanning includes all git-ignored files and follows any
symlinks. When any directory changes on disk, Zed recursively rescans it
in order to keep its in-memory representation up-to-date. When
collaborating, all of these files are replicated to all guests.
Right now, there are some performance problems associated with the
maintenance of this filesystem state:
* For various reasons, some projects contain symlinks that point out to
large folders like `$HOME`, which itself contains many symlinks that
point to the same large directory. When these projects are opened, the
worktree scans endlessly, using more and more memory.
* Some git-ignored directories (like `target` in a rust project) contain
*many* more files than are actually tracked in the git repository. These
files often change as a result of saving, (e.g. because the compiler
runs). Maintaining in memory all of these paths isn't useful to the
user, and causes significant CPU usage on every save. Most importantly,
when collaborating sending all of these changes to guests can be slow,
and can delay all other RPC messages.
### Change
This PR changes the worktree's filesystem-scanning logic to be *lazy*
about scanning two types of directories:
* git ignored directories
* "external" directories (those that are canonically located outside of
the worktree root, but accessed via symlinks)
The laziness works as follows. When, during a recursive scan, a
directory is found that falls into one of the above 2 categories, that
directory is marked as "unloaded". The directory might later be scanned,
if some explicit operation is performed within it (like opening a
buffer, or creating a file), if any collaborator expands that directory
in their project panel, or if an LSP requests that it be watched.
### Results
When collaborating on the `zed` folder:
| metric | before | after |
|-------|--------|------|
| # `worktree_entries` in collab db initially | 154,763 | 77,679 |
| # `worktree_entries` in collab db after 5 saves | 181,952 | 77,679
(nothing new to scan) |
| app memory footprint (host) | 260MB | 228.5 MB |
The db thing is a win, because reading and writing to the
`worktree_entries` table is one of the most expensive thing that the
`collab` server does.
There's also generally lower background CPU usage after every save,
because we don't need to recursively rescan directories inside of
`target`.
### Limitations
We still end up scanning some unnecessary directories (like
`target/debug/build/zed-b612db829aeac16e/out`) because the LSP instructs
us to watch those.
### To do:
* [x] Expand parent directories of any path opened via LSP
* [x] Avoid creating orphaned entries when FS events happen inside of
unscanned directories
* [x] Scan any newly-non-ignored directories after gitignore changes
* [x] Emit correct events for newly-discovered paths when expanding dirs
* [x] GC the set of expanded directory ids when dirs are removed
* [x] Don't include "external" entries in file-finder
* [x] Expand any directories watched by LSP
* [ ] manual testing and profiling
### Release Notes:
- Fixed a bug where Zed would use excessive memory when a project folder
contained symlinks pointing to directories outside of the project.
- Reduced Zed's memory and CPU usage when working in folders containing
many git-ignored files.
After #2641 we noticed that scrolling didn't take a count parameter, and
a few other issues with the way that we calculated the distance to
scroll.
Release Notes:
- Improved distance calculations for page-up/page-down
- vim: Allow counts to work with scrolling shortcuts.
Fixes: zed-industries/community#1690
I'm not sure this is the correct way to fix this...
* A simpler approach would be to just say `!showing_code_actions` in the
binding file (as `showing_completions` can only happen in insert mode -
and `VimControl` will be false). This seemed a little error prone if
more menus were added in the future.
* A more complicated approach would be to copy the way this is done from
the MouseContextMenu, which registers its own keyboard shortcuts, and as
such uses those when it's open. This seems "more correct", but is a
major refactoring for a very small reward.
Release Notes:
- vim: Fix code actions menu
([#1690](https://github.com/zed-industries/community/issues/1690))
After #2641 we noticed that scrolling didn't take a count parameter.
The PageDown/PageUp logic was also broken by an additional -1 (for both
vim mode and not).
We want to add installation_id to the panic events so that we can easily
know if multiple panics are coming from the same person or different
people. 5 panics from one person isn't as bad as 5 panics from 5 people.
[zed.dev pr](https://github.com/zed-industries/zed.dev/pull/343)
Release Notes:
- N/A
[[PR Description]]
When we updated the way themes were built it looks like we accidentally
gave `Atelier Forest Light` the wrong `ThemeAppearance`. This PR fixes
that.
Release Notes:
- Fixed the `Atelier Forest Light` theme, which was incorrectly set as a
dark theme.
The test was testing pretty straightforward logic, but for some strange reason
it was failing on CI (but passed locally). I think it's fine to delete it and
make progress, if zooming regresses we'll find out pretty quickly.
To turn any struct into a composite element, you can implement a render
method with the following signature:
```rs
fn render<V: View>(&mut self, view: &mut V, cx: &mut ViewContext<V>) -> AnyElement<V>;
```
Then add #[derive(Element)] to the struct definition.
This will make it easier to introduce higher-level components that are
expressed in terms of other elements. Instead of calling functions that
return elements, we can now make any struct into an element fairly
easily. The advantage is that we can use method chaining to express
optional state on these components, and they blend in better with other
elements.
cc @mikayla-maki @osiewicz @iamnbutler
Release Notes:
- N/A
To turn any struct into a composite element, you can implement a render method
with the following signature:
fn render<V: View>(&mut self, view: &mut V, cx: &mut ViewContext<V>) -> AnyElement<V>;
Then add #[derive(Element)] to the struct definition.
This will make it easier to introduce higher-level components that are expressed in
terms of other elements.
These adapters have indicated some broader reason to the user why
they cannot be started, don't waste time/bandwidth attempting to
validate and reinstall them
Before this change code could not distinguish between a user providing a
count of 1 and no count at all.
Fixes: zed-industries/community#710
Release Notes:
- 1G now correctly goes to the first line in vim mode
([710](zed-industries/community#710))
Don't wait until populating that directory entry, for two reasons:
* In the case of submodules, .git is not a directory
* We don't eagerly populate .git directories, since their contents
are automatically ignored.
This PR adds JSON Schema definitions into GPUI's styles and adds a tool
for generating typescript types from these schema definitions.
Release Notes:
- N/A
Release Notes:
- In vim mode you can now use count modifiers with arrow keys (`5<down>`
will go down 5 lines).
- In vim mode `ctrl+]` and `ctrl+o` now work to go to definition and
back
🚧 We have a couple more refinements to this to add 🚧
Release Notes:
You can now join lines with `ctrl-j` or `shift-J` in Vim normal mode.
🍐'd with @ConradIrwin
This is a part of the intensity driven theme rewrite.
It introduces the `toggle` and `interactive` helper functions to build
Toggle<T> and Interactive<T> styles for interactive elements in the
theme.
This PR also removes the `theme_testbench` crate and related actions.
Huge thanks to @osiewicz and @mikayla-maki for pushing this forward 🙏🏽
Release Notes:
- Updated the style of many interactive elements.
Improves latency for big inlay hints LSP responses for ~8k line files.
Before, the CPU usage sample for editing a single line inside
`edirot.rs` file in Zed contained serde inside the main thread traces:
<img width="1728" alt="Screenshot 2023-06-21 at 00 33 23"
src="https://github.com/zed-industries/zed/assets/2690773/d9789efe-8055-487f-bbe7-8beb49605bcb">
Release Notes:
- N/A
Since we don't want tabs, I think it would be better to render the toolbar
for ourselves directly and handle switching between conversations.
Co-Authored-By: Julia Risley <julia@zed.dev>
I finally got fed up with being unable to copy error messages. This adds
a click target and tooltip to f8-style diagnostics that copies their
text on click.
Release Notes:
- Added the ability to copy under-line diagnostic errors on click
Fix mislocation of caller query in detach_and_log_error
Fix incorrect wording on livekit integration
Add share_mic action for manually enabling the microphone
Make mic sharing wait until the room has been fully established
This avoids a high cost which appears to be the system rasterizing the
cursor every time we call this, fixes a slowdown when scrolling rapidly
while mouse motion continually attempted to assign the style
Co-Authored-By: Antonio Scandurra <me@as-cii.com>
Z-2357
I've found a crate that handles both comments and trailing commas in
JSON. It is a fork of `serde_json`, but it is maintained & up-to-date.
Sadly RawValue seems to not play nicely with it; I've ran into
deserialisation issues around use of RawValue. For this PR I've migrated
to `Value` API.
Obviously this is just a point of discussion, not something I'd merge
straight away. There may be better solutions to this particular problem.
I've also noticed that `serde_json_lenient` does not handle trailing
commas after bindings array. I'm not sure how big of an issue that is.
Release Notes:
- Improved handling of trailing commas in settings files.
[#1322](https://github.com/zed-industries/community/issues/1322)
Still need to implement loading / listing.
I'd really be rather write operations to a database. Maybe we
should be auto-saving? Integrating with panes? I just did
the simple thing for now.
I'd like to follow up to allow roles to be cycled for the selected range
and support multi-cursors, but this is a start and contains a
refactoring, so going to merge.
Release Notes:
- Added the ability to cycle roles in the assistant with `ctrl-r`
When multiple panics occur at the same time (usually because one thread
panics, and another thread joins it), multiple panic JSON objects can
get written to the same panic file. The resulting file won't be valid
JSON.
This PR addresses that problem via two changes:
* Format panic files as single-line JSON objects
* When a panic file isn't valid JSON, try taking the first line
In the future, we could try combining all of the backtraces, but for
now, I just want to avoid a problem of not reporting a panic at all.
Release Notes:
- Fixed a problem with Zed's internal crash reporting.
This PR updates some dock behaviors. Now the toggle-dock commands
(cmd-j/b/r) also toggle focus. This also adds zoom serialization to the
docks.
Release Notes:
- Bug fix: Toggle dock commands (cmd-j/b/r) now move focus
- Bug fix: Dock zoom is now restored with the rest of the workspace
This fixes a regression introduced in
https://github.com/zed-industries/zed/pull/2560, where panic reports did
not include backtraces. The problem was that in that PR, I assumed we
could retrieve file paths for symbols in our backtraces. But actually,
that functionality only works when the app is built locally, and a
`.dSYM` file can be magically found by the OS. We don't ship those dSYM
files with Zed, so panic symbols do not have file paths available.
Panic backtraces will still be more useful and less noisy than before
though: we will strip out frames for which we don't have symbol names,
and remove leading panic-handling stack frames from the backtraces.
Release Notes:
- N/A
This PR adds internal docs directly to the codebase.
There are a few goals here:
- Make it easier to document our internal processes
- Put the documentation where people already are
- Allow changes to be in sync with PRs
- Make it easier for people coming in to discover our docs and onboard
faster.
With 2 more people joining us in the next few weeks it would be great to
get these up.
Release Notes:
- N/A (No public facing changes)
This completes the bundle changes that will be needed to access voice,
as well as adds permissions for accessing other MacOS services, the
camera, and the necessary permissions for plugins. This was developed by
combining the entitlements of iTerm and VSCode, cross-referenced with
the entitlements of Firefox.
Release Notes:
- Fixed a bug in enabling authorization for macOS services (preview
only)
This updates our dependency on cargo-bundle to point at a fork I've
created that adjusts the latest version's behavior to work for our use
case. This PR also updates the Zed Info.plist to include the following
keys:
- NSSystemAdministrationUsageDescription
- NSAppleEventsUsageDescription
- NSBluetoothAlwaysUsageDescription
- NSCalendarsUsageDescription
- NSCameraUsageDescription
- NSContactsUsageDescription
- NSLocationAlwaysUsageDescription
- NSLocationUsageDescription
- NSLocationWhenInUseUsageDescription
- NSMicrophoneUsageDescription
- NSSpeechRecognitionUsageDescription
- NSRemindersUsageDescription
Release Notes:
- Bug fix: Allow applications run via the Zed terminal to request
authorization for macOS services (e.g. microphone and camera)
Closes
https://linear.app/zed-industries/issue/Z-2368/use-a-different-icon-for-the-assistant-panel
Closes
https://linear.app/zed-industries/issue/Z-2363/ship-the-assistant-only-on-preview
Closes
https://linear.app/zed-industries/issue/Z-2331/scrolling-makes-it-hard-to-read
Closes
https://linear.app/zed-industries/issue/Z-2306/allow-undo-and-collaboration-in-assistant
This pull request is a significant overhaul of the assistant panel,
which now uses a simple `Buffer` as opposed to a `MultiBuffer` to show
messages. Specifically, we track the start of each message with an
anchor located right after the newline (or `Anchor::MIN` for the first
message). When the anchor becomes invalid (that is, the newline is
deleted), we merge the message with the preceding ones. Crucially,
messages don't actually get deleted so that, if the newline anchor
becomes valid again (such as when undoing/redoing), we can restore the
messages as well.
As part of this overhaul, we are also improving the scrolling behavior
to maintain the viewport stable only when editing or moving the cursor,
but otherwise leave the scroll position unchanged when manually
scrolling up or down.
Note that with these changes, we are limiting access to the assistant to
users on preview (and dev), as we want to polish the behavior a little
more before shipping to the general public. Users on stable will still
be able to see the default settings/keybindings of the assistant, but I
think that's okay, as they won't be able to do anything with them.
Release Notes:
- Added support for undo/redo in the assistant (preview-only)
- Improved the scrolling behavior of the assistant when it was
generating responses. Now Zed will keep the viewport stable only when
editing or moving the cursor, but otherwise leave the scroll position
unchanged when manually scrolling up or down (preview-only)
- Changed the icon of the assistant panel (preview-only)
**Note for @JosephTLyons: given that we're feature flagging this, let's
make sure things on stable look reasonable and work correctly. Things to
look out for: ensure a stock installation works, changing the settings
on stable works, changing the keybinding on stable works.**
The worktree has a lot of tests. Let's reduce their indentation, and
make it easier to navigate to them. We deserve it.
Just doing this in its own PR, to minimize conflicts.
Release Notes:
- N/A
Per @JosephTLyons request I've added a language setting for comment
continuations.
Release Notes:
- Added a language setting for comment continuations.
This PR adds a syntax tree view, which lets you view the syntax tree of
any layer in the active editor's `SyntaxMap`.
This view uses some new APIs that I added to Tree-sitter, which allow us
to efficiently render the syntax tree using a `UniformList`. Tree-sitter
PR: https://github.com/tree-sitter/tree-sitter/pull/2316

Release Notes:
- Added a *syntax tree view* that shows Zed's internal syntax tree(s)
for the active editor. You can open it running the `debug: open syntax
tree view` command from the command palette.
Some user's don't have their system clocks configured right and we are
seeing events 10 years into the future. I'm stripping out the code that
adds time via the client and am adding it in on zed.dev. We will lose a
little accuracy, as the time will be when the batch hits the server, but
I think its negligible (currently, batches send up every 30 seconds, at
the max) and worth it to protect our data from user's who wonkily dont
set care about their system time.
- https://github.com/zed-industries/zed.dev/pull/332
Release Notes:
- N/A
This PR adds the ability to export tokens for each theme.
You can export tokens by:
1. `cd styles`
2. `npm run build-tokens`
3. Tokens will be output in the target folder (`styles/target`)
The tokens match the ColorScheme object. In the future we may also
export tokens for our styleTrees.
Release Notes:
- N/A (No public facing changes)
---
TODO:
- [x] Generate Token Studio theme index file
- [x] ColorScheme
- [x] name:
- [x] isLight
- [x] lowest
- [x] middle
- [x] highest
- [x] popoverShadow
- [x] modalShadow
- [x] players
- [x] syntax
Also, I noticed errors in the logs of the Elixir LSP that we were
sending `goToTypeDefinition` requests, which that server does not
support. We now respect that server capability.
Release Notes:
- N/A
In debugging what's going on with the Elixir language server, there was
some interesting content in the server's logs (sent to the app via the
`window/logMessage` LSP endpoint). I decided to invest in making
language server issues easier to debug by exposing these `logMessage`
contents in the app.
Also, improve the UI of the view slightly:
* Select one of the servers by default (instead of "no server selected")
* Make it clearer that the menu is clickable
We removed our Figma Tokens/Tokens Studio export a while back when we
moved to the theme to ColorSchemes. I'd like to get back to exporting
these so we can be working from up to date themes in Figma, especially
with the large amount of UI design work we'll be doing in the next few
weeks on channels.
This PR adds some basic plumbing to start working through the
theme/colorScheme and export the parts as tokens.
I also discovered that Tokens Studio now publishes their types, so we
can use them directly rather than writing our own:
https://github.com/tokens-studio/types
Pulled those in and started connecting them as well.
Running `npm run build-tokens` will export the tokens for each theme to
`styles/target/tokens`.
Currently only a few element's tokens are exported, will expand this
further as time permits.
Release Notes:
- N/A (No public facing changes)
We've accumulated a number of miscellaneous folders in the styles
folder. Simplifying these down to `theme`, `themes` and `styleTrees`.
Release Notes:
- N/A (No public facing changes)
This rewrites and simplifies the git status system by attaching the git
status to each individual entry. This also improves the git testing
infrastructure to cover more cases and be more accurate to how file
events actually occur. This also fixes several other bugs in the worktree and the buffer, and stops any randomly generated actions from happening inside a `.git` folder. Hopefully, we can undo this change soon once our randomized testing is more robust.
Release Notes:
- Will require a DB migration
TODO:
- [x] Pass randomized tests
- [x] Get ready for merging
Fixes Z-1618. In the current state, this only works for line comments
such as `//` (and whatever's set in `{language}.toml` as a
line_comment).
Release Notes:
- Comments are now extended on new line.
This pull request introduces a new assistant panel to Zed that lets
users interact with OpenAI using their API key:

After setting the key up, it will be saved to the keychain and
automatically loaded the next time the assistant panel is opened. The
key can be reset using `assistant: reset key`.

From there, users can type messages in a multi-buffer and hit
`cmd-enter` (`assistant: assist`) to stream assistant responses using
the OpenAI API. Responses can be canceled by hitting `escape`.

Users can quote a selection from the active editor by hitting `cmd->`
(`assistant: quote selection`), which will embed the selected piece of
text in a Markdown fenced code block. Conversations with the assistant
are ephemeral at the moment, but can be easily copy/pasted:

Release Notes:
- Added a new assistant panel feature that enables interacting with
OpenAI using an API key. This replaces the previous experimental `ai:
assist` command that would work on any buffer. The experience is similar
to the one offered by ChatGPT with the added ability to edit, delete or
enhance previous messages. When hitting `cmd-enter`, the assistant will
start streaming responses from OpenAI. A response stream can be canceled
using `escape`. Moreover, the active editor's selection can be quoted in
the assistant panel using `cmd->`, which will automatically embed the
selected piece of text in a Markdown fenced code block.
Fixes
https://linear.app/zed-industries/issue/Z-2208/outline-view-doesnt-differentiate-between-overloaded-functions
Fixes
https://linear.app/zed-industries/issue/Z-2205/elixir-syntax-highlighting-not-working-properly-for-doc-attributes-and
This PR improves syntax highlighting and outline view in Elixir. It's
common to overload elixir functions, with many different versions of the
function for different patterns of parameters, so I updated the outline
view to show functions' parameters in Elixir. But if we showed functions
the same way in the *breadcrumbs*, it would take up too much space.
So I added a new capture in languages' `outline` queries called
`@context.extra`, which is included in the outline view, but not in
breadcrumbs.
Release Notes:
- Improved syntax highlighting of doc attributes and special macros in
Elixir
- Updated the outline view in Elixir to display function parameters, to
allow differentiating between function overloads.
This PR adds live reloading of Tree-sitter queries when running in debug
mode, similar to what we do for the themes. This way, you can change a
highlighting query or an outline query, and immediately see the result
in the app.
Release Notes:
- N/A
Per conversation with Antonio, I've suggested enabling full LTO; right
now we use a crate-local ThinLTO, which does not inline function calls
across crates.
| Configuration | Current main (788f97e) | Thin LTO | Full LTO |
|------------------|------------------------|-----------|-----------|
| Size in bytes | 158806721 | 155868753 | 111115553 |
| % of `main` size | 100% | 98.14% | 69.96% |
| Size in bytes (no debug info) | 129186657 | 127942929 | 108281345 |
Previously, all static assets were embedded into Zed using a single
`RustEmbed` struct called `Assets`, which was in a crate of its own,
also called `assets`. Many crates depended on this crate. The problem
with this situation was that changing *any* static asset file caused
almost every crate in the codebase to need recompilation (because of
recursive dependencies on `assets`).
Now, most of the assets are embedded only into the top-level `zed`
crate. A few assets (such as settings JSON files and AI prompts) are
needed in lower-level crates such as `settings` and `ai`. For these,
I've created separate `RustEmbed` structs in those specific crates,
which embed those specific sub-folders of `assets`.
The result is that now, when you change a theme file, the only crate
that needs to recompile is `zed`.
Release Notes:
- N/A
Previously, project-specific settings were not discoverable. This PR
adds a `Zed > Preferences > Local Settings` application menu command
that creates a `.zed/settings.json` at the root of your current
worktree. This command works in both local and remote projects.
Limitations:
* Currently, if you have an empty project open, the command just shows a
notification that there are no folders open.
* The JSON-schema-based autocomplete is the same in local settings files
as in your main settings, even though not all settings can be locally
customized.
Release Notes:
- Added an application menu command - `Zed > Preferences > Local
Settings` for creating a folder-specific settings file.
Following https://github.com/zed-industries/zed/pull/2559 the project
panel entries become pretty wide again. This PR tries to mitigate that
and just make some general improvements to visual density in the project
panel.
- Reduces padding around items
- Removes top margin
- Slightly reduces the height of each item
- Fixes an issue where ignored files had the wrong color chevron
Release Notes:
- Improved density of the project panel and tidied up some visual
issues.
Problem is, I'm trying to trust the excerpt id of the selection head,
but it's a sentinel value and not the actual excerpt id of the message.
I think we probably need to resolve to offsets instead.
The `start-local-collaboration` script opens two instances of Zed, each
logged in as a different user, and each one taking up half of the
screen. But previously, when joining a remote project as one of the
collaborators, that newly-opened window would be full screen.
Now, each instance of Zed keeps *all* of its windows on its half of the
screen. This is implemented by respecting the `ZED_WINDOW_{SIZE,BOUNDS}`
env vars, even when joining *remote* projects.
Release Notes:
- N/A
Fixes
https://linear.app/zed-industries/issue/Z-1511/thread-main-panicked-at-assertion-failed-left-==-right-left-local-0-1
Previously, when exchanging messages about buffers with a copilot
language server, we identified buffers using their **remote id**. This
caused problems when there were multiple projects open, where one or
more were remote, because buffers' remote ids are only unique within a
given project.
When you have multiple projects open, and one or more of the projects is
remote, it's pretty easy to have two buffers open with the same remote
id. In my testing, when this happened, copilot would stop working in
both buffers. But I believe that depending on the editing patterns that
occur in the two buffers, it could cause the crash reported in the
Linear issue above.
This PR changes our copilot logic to use buffers' local handle ids for
identifying them. This fixed the problems I was able to reproduce when
using copilot in both remote and local projects.
Release Notes:
- Fixed a crash that would sometimes occur when editing buffers after
having collaborated on a remote project.
This PR adds codegen from rustc to track the file and line number of
calls to `log_err()`. I haven't noticed much longer compile times on my
machine, and looking at the
[implementation](https://rustc-dev-guide.rust-lang.org/backend/implicit-caller-location.html)
it essentially adds an extra argument and secret reference pass.
However, this will show a lot more data in our logs on user machines.
Requesting review from @ForLoveOfCats, who usually knows a bunch about
this kind of thing :)
I did \*something\* Friday afternoon which changed something about my
system SDK to break the `media` crate's bindings generation. Some of the
types and consts were not being generated despite being able to prove
that they exist in the source, such as when running the header through
the preprocessor myself and feeding that through the generator. Updating
my OS, XCode, command line tools, and reinstalling Rust as well as
working from fresh clones of the repo had no effect.
Updating rust-bindgen resolved the issue and downgrading the version
back to the original version caused the issue to reappear. I'm still not
sure what happened to change the SDK but at this point with being able
to build the project again I'm not going to look a gift horse in the
mouth.
Release Notes:
- N/A
This was causing IME input to be drawn in the wrong place when there
were splits or panels in the window.
Release Notes:
- Fixed a bug that was causing IME input to sometimes be rendered in the
wrong position.
Quite literally just ran `typos --write-changes` from
https://crates.io/crates/typos. Its pretty impressive - wonder if we
should run it on CI.
Release Notes:
- N/A
* Add an `identifying_backtrace` field that only contains symbols in
*our* own codebase, which can be used for better deduplication.
* In the main backtrace, include file and line numbers for all symbols
in our codebase
* Exclude any stack frames within the panic handling/hooking system
itself, so that the top line of the backtrace is where the panic
originated in our codebase.
This should improve our panic deduplication, and also make panic reports
a bit more readable.
example:
```
{
"thread": "main",
"payload": "wtf",
"location_data": {
"file": "crates/zed/src/zed.rs",
"line": 459
},
"backtrace": [
"zed::open_log_file::{{closure}}::{{closure}}::{{closure}}",
" crates/zed/src/zed.rs:459",
"gpui::app::AppContext::spawn_internal::{{closure}}",
" crates/gpui/src/app.rs:2073",
"gpui::executor::any_local_future::{{closure}}",
" crates/gpui/src/executor.rs:1026",
"<core::pin::Pin<P> as core::future::future::Future>::poll",
"<async_task::runnable::spawn_local::Checked<F> as core::future::future::Future>::poll",
"async_task::raw::RawTask<F,T,S>::run",
"async_task::runnable::Runnable::run",
"<gpui::platform::mac::dispatcher::Dispatcher as gpui::platform::Dispatcher>::run_on_main_thread::trampoline",
" crates/gpui/src/platform/mac/dispatcher.rs:40",
"<() as objc::message::MessageArguments>::invoke",
"objc::message::platform::send_unverified",
"objc::message::send_message",
"<gpui::platform::mac::platform::MacForegroundPlatform as gpui::platform::ForegroundPlatform>::run",
" crates/gpui/src/platform/mac/platform.rs:366",
"gpui::app::App::run",
" crates/gpui/src/app.rs:251",
"Zed::main",
" crates/zed/src/main.rs:118",
"core::ops::function::FnOnce::call_once",
"std::sys_common::backtrace::__rust_begin_short_backtrace",
"std::rt::lang_start::{{closure}}",
"core::ops::function::impls::<impl core::ops::function::FnOnce<A> for &F>::call_once",
"std::rt::lang_start"
],
"release_channel": "dev",
"os_name": "macOS",
"os_version": "12.6.1",
"architecture": "aarch64",
"panicked_on": 1685734744050,
"identifying_backtrace": [
"zed::open_log_file::{{closure}}::{{closure}}::{{closure}}",
"gpui::app::AppContext::spawn_internal::{{closure}}",
"gpui::executor::any_local_future::{{closure}}",
"<gpui::platform::mac::dispatcher::Dispatcher as gpui::platform::Dispatcher>::run_on_main_thread::trampoline",
"<gpui::platform::mac::platform::MacForegroundPlatform as gpui::platform::ForegroundPlatform>::run",
"gpui::app::App::run",
"Zed::main"
]
}
```
Release Notes:
N/A
* Add an 'identifying_backtrace' field that only contains symbols in our
codebase, which can be used for better deduplication.
* In the main backtrace, include file and line numbers for all symbols
in our codebase
For whatever reason, the optimizations of panes and workspace have
caused the terminal to notify less often then it should. This PR fixes
that oversight.
The files will still get created if the user opens their settings and
saves, otherwise everything will transparently work
Release Notes:
- Fixed an issue where a missing settings file would cause a hang on
startup
([#1590](https://github.com/zed-industries/community/issues/1590)).
The files will still get created if the user opens their settings and
saves, otherwise everything will transparently work
Co-Authored-By: Antonio Scandurra <me@as-cii.com>
Co-Authored-By: Max Brunsfeld <max@zed.dev>
This PR allows you to customize Zed's settings within a particular
folder by creating a `.zed/settings.json` file within that folder.
Todo
* [x] respect folder-specific settings for local projects
* [x] respect folder-specific settings in remote projects
* [x] pass a path when retrieving editor/language settings
* [x] pass a path when retrieving copilot settings
* [ ] update the `Setting` trait to make it clear which types of
settings are locally overridable
Release Notes:
* Added support for folder-specific settings. You can customize Zed's
settings within a particular folder by creating a `.zed` directory and a
`.zed/settings.json` file within that folder.
Release Notes:
* Fixed the behavior of panel buttons when their panel is open. Show the
key binding for closing the containing dock (preview only).
* Improved the styling of zoomed panels and panes, giving them a
stronger border, and color the zoom button with an "active" style
(preview only).
* Improved the stability of panels' zoom state. Close their dock instead
of resetting their zoom state when dismissing panels to reveal other
items (preview only).
This PR updates the dock key bindings according to the following model:
There are three bits:
Visible: Opened / closed.
Focus: Panel focused / center focused.
Zoom: Zoomed / Not zoomed.
Each of these variables is 'sticky' in that they won't effect each other
unless they need to. 'Zooming' a panel conceptually merges the visible
and focus bits.
cmd-shift-j/b/r have all been removed.
cmd-j/b/r have been updated to mean 'toggle visibility of a certain
dock', firing them should *always* reveal the panel to you (where you
last left it), or hide it, without moving focus (unless the focused
element is invisible). This means that, when the terminal panel is
zoomed, cmd-j has the same effect as ctrl-`
ctrl-` and cmd-shift-e now toggle a panel's focus, without updating the
zoom state of a panel. Toggling the focus of a zoomed panel causes it to
automatically hide itself, without losing the zoom bit.
When focused or made visible, panels which cannot be zoomed
automatically unzoom everything else so as to preserve user intent of
'show me this panel' and 'everything stays where it is if I don't take
an action'
Release Notes:
- cmd-shift-j/b/r have been removed. (preview only)
- cmd-j/b/r unconditionally show or hide their associated dock,
respecting zoom settings. (preview only)
- ctrl-` and cmd-shift-e now retain zoom state. (preview only)
- Fixed a bug where terminal dock tab would always be in the active
state (preview only)
- Fixed a bug where terminals would not always open in the terminal
panel
- Changed the look of zoomed panels to fill more of the screen (preview
only)
Rather than assuming a specific family exists, try a set of specific
names and if they fail, just grab any old font that the system reports
as existing
Co-Authored-By: Nathan Sobo <nathan@zed.dev>
This pull request simplifies the `Pane` struct by replacing methods like
`Pane::add_item` that would previously take a `&mut Workspace` with
methods that take a `&mut self`. When access to the workspace is needed,
we now either emit an event from the `Pane` or directly move the method
to the `Workspace` struct.
I was having fun, but it was too much. It makes the assistant too
annoying to be useful.
Release Notes:
Use `N/A` in this section if this item should be skipped in the release
notes.
Add release note lines here:
* Dialed back the system prompt on the assistant.
* ESC (project_search::ToggleFocus) toggles focus from include/exclude
fields to the editor
* Cmd+Shift+F (workspace::NewSearch) can be triggered from the editor,
and moves focus to the query editor
Release Notes:
* Improved project search panel shortcut handling, allowing more actions
to trigger from panel elements
* ESC (project_search::ToggleFocus) toggles focus from
include/exclude fields to the editor
* Cmd+Shift+F (workspace::NewSearch) can be triggered from the editor,
and moves focus to the query editor
This PR makes the worktree's change events more useful in a few ways:
* The changes are now described by a cheaply clone-able collection, so
that they can be used in background tasks. Right now, I'm using a simple
Arc slice.
* The `UpdatedEntries` event now captures not only changes due to FS
changes, but also newly-loaded paths that are discovered during the
initial scan.
* The `UpdatedGitRepositories` event now includes repositories whose
work-dir changed but git dir didn't change. A boolean flag is included,
to indicate whether the git content changed.
* The `UpdatedEntries` and `UpdatedGitRepositories` events are now
*used* to compute the worktree's `UpdateWorktree` messages, used to sync
changes to guests. This unifies two closely-related code paths, and
makes the host more efficient when collaborating, because the
`UpdateWorktree` message computation used to require walking the entire
`entries` tree on every FS change.
We don't need to look-up change types by an arbitrary key, but we
do need to iterate it. It would also be useful to be able to
cheaply clone the changes, to use them in a background task.
Each message is represented as a multibuffer excerpt to allow for
fluid editing of the conversation transcript.
Co-Authored-By: Antonio Scandurra <antonio@zed.dev>
This took so much longer than I wanted, so glad to finally be rid of
this
Release Notes:
* Improved performance when editing many git-tracked files in a
multi-buffer at the same time
This avoids an issue where in a many-buffer multi-buffer, each modified
buffer could complete its recalc independently, causing a cascade of
repeated notifies
Now all recalcs started at the same time must complete before
A: Starting another recalc pass
B: The master notify occurring
Each buffer can still show its new diff if something else triggers it
to notify earlier, this is desirable and does not have the same negative
effects as the notify cascade as those re-layouts would need to happen
anyway
Co-Authored-By: Max Brunsfeld <max@zed.dev>
For some reason the yellow I used for the modified color in light themes
was really light
.
Release Notes:
* Improved the contrast of diff modified color in the editor in light
themes.
This adds basic ability to interact with OpenAI inside a buffer.
Release Notes:
* If`OPENAI_API_KEY` is defined in your environment, you can now run the
`ai: assist` command to pass the text of your current buffer to GPT-4.
If you're editing a file with a `.zmd` extension, you can also invoke
the model with `cmd-enter.` (preview-only)
Drop dependency on tokio introduced by async-openai and do it ourselves.
The approach I'm taking of replacing instead of appending is causing issues. Need to just append.
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.