Release Notes:
- Fixed a bug that would prevent the agent from working over SSH.
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Richard Feldman <oss@rtfeldman.com>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Cole Miller <m@cole-miller.net>
Adds a new `agent.model_parameters` setting that allows the user to
specify a custom temperature for a provider AND/OR model:
```json5
"model_parameters": [
// To set parameters for all requests to OpenAI models:
{
"provider": "openai",
"temperature": 0.5
},
// To set parameters for all requests in general:
{
"temperature": 0
},
// To set parameters for a specific provider and model:
{
"provider": "zed.dev",
"model": "claude-3-7-sonnet-latest",
"temperature": 1.0
}
],
```
Release Notes:
- agent: Allow customizing temperature by provider/model
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Marshall Bowers <git@maxdeviant.com>
When using the agent with a project shared by a collaborator, rules file
loading didn't work as it was trying to read from the client's
filesystem
Release Notes:
- Fixed rules file loading when using the agent with a project shared by
a collaborator.
Things this doesn't currently handle:
- [x] ~testing~
- ~we really need an snapshot test that takes a vscode settings file
with all options that we support, and verifies the zed settings file you
get from importing it, both from an empty starting file or one with lots
of conflicts. that way we can open said vscode settings file in vscode
to ensure that those options all still exist in the future.~
- Discussed this, we don't think this will meaningfully protect us from
future failures, and we will just do this as a manual validation step
before merging this PR. Any imports that have meaningfully complex
translation steps should still be tested.
- [x] confirmation (right now it just clobbers your settings file
silently)
- it'd be really cool if we could show a diff multibuffer of your
current settings with the result of the vscode import and let you pick
"hunks" to keep, but that's probably too much effort for this feature,
especially given that we expect most of the people using it to have an
empty/barebones zed config when they run the import.
- [x] ~UI in the "welcome" page~
- we're planning on redoing our welcome/walkthrough experience anyways,
but in the meantime it'd be nice to conditionally show a button there if
we see a user level vscode config
- we'll add it to the UI when we land the new walkthrough experience,
for now it'll be accessible through the action
- [ ] project-specific settings
- handling translation of `.vscode/settings.json` or `.code-workspace`
settings to `.zed/settings.json` will come in a future PR, along with UI
to prompt the user for those actions when opening a project with local
vscode settings for the first time
- [ ] extension settings
- we probably want to do a best-effort pass of popular extensions like
vim and git lens
- it's also possible to look for installed/enabled extensions with `code
--list-extensions`, but we'd have to maintain some sort of mapping of
those to our settings and/or extensions
- [ ] LSP settings
- these are tricky without access to the json schemas for various
language server extensions. we could probably manage to do translations
for a couple popular languages and avoid solving it in the general case.
- [ ] platform specific settings (`[macos].blah`)
- this is blocked on #16392 which I'm hoping to address soon
- [ ] language specific settings (`[rust].foo`)
- totally doable, just haven't gotten to it yet
~We may want to put this behind some kind of flag and/or not land it
until some of the above issues are addressed, given that we expect
people to only run this importer once there's an incentive to get it
right the first time. Maybe we land it alongside a keymap importer so
you don't have to go through separate imports for those?~
We are gonna land this as-is, all these unchecked items at the bottom
will be addressed in followup PRs, so maybe don't run the importer for
now if you have a large and complex VsCode settings file you'd like to
import.
Release Notes:
- Added a VSCode settings importer, available via a
`zed::ImportVsCodeSettings` action
---------
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
Co-authored-by: Kirill Bulatov <kirill@zed.dev>
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
Co-authored-by: Marshall Bowers <git@maxdeviant.com>
Now that we've established a proper eval in tree, this PR is reboots of
our agent loop back to a set of minimal tools and simpler prompts. We
should aim to get this branch feeling subjectively competitive with
what's on main and then merge it, and build from there.
Let's invest in our eval and use it to drive better performance of the
agent loop. How you can help: Pick an example, and then make the outcome
faster or better. It's fine to even use your own subjective judgment, as
our evaluation criteria likely need tuning as well at this point. Focus
on making the agent work better in your own subjective experience first.
Let's focus on simple/practical improvements to make this thing work
better, then determine how we can craft our judgment criteria to lock
those improvements in.
Release Notes:
- N/A
---------
Co-authored-by: Max <max@zed.dev>
Co-authored-by: Antonio <antonio@zed.dev>
Co-authored-by: Agus <agus@zed.dev>
Co-authored-by: Richard <richard@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Michael Sloan <mgsloan@gmail.com>
Release Notes:
- Added the ability to copy external files into remote projects by
dragging them onto the project panel.
---------
Co-authored-by: Peter Tripp <petertripp@gmail.com>
The implementation of commondir discovery in #27885 was wrong, most
significantly for submodules but also for worktrees in rarer cases. The
correct procedure, implemented in this PR, is:
> If `.git` is a file, look at the `gitdir` it points to. If that
directory has a file called `commondir`, read that file to find the
commondir. (This is what happens for worktrees.) Otherwise, the
commondir is the same as the gitdir. (This is what happens for
submodules.)
Release Notes:
- N/A
This fixes an issue where tasks in `.vscode/tasks.json` weren't being
loaded at startup of a project
Closes#28494
Release Notes:
- Tasks are now loaded from local `.vscode/tasks.json` files even if
they are `.gitignore`d
Avoids building a whole git2 repository object at the worktree layer
just to watch some additional paths.
- [x] Tidy up names of the various paths
- [x] Tests for worktrees and submodules
Release Notes:
- N/A
We cannot compare versions and anchors between different `Buffer`s with
different `BufferId`s.
Release Notes:
- Fixed Zed panicking on editor reopen
Co-authored-by: Conrad Irwin <conrad@zed.dev>
This adds a "workspace-hack" crate, see
[mozilla's](https://hg.mozilla.org/mozilla-central/file/3a265fdc9f33e5946f0ca0a04af73acd7e6d1a39/build/workspace-hack/Cargo.toml#l7)
for a concise explanation of why this is useful. For us in practice this
means that if I were to run all the tests (`cargo nextest r
--workspace`) and then `cargo r`, all the deps from the previous cargo
command will be reused. Before this PR it would rebuild many deps due to
resolving different sets of features for them. For me this frequently
caused long rebuilds when things "should" already be cached.
To avoid manually maintaining our workspace-hack crate, we will use
[cargo hakari](https://docs.rs/cargo-hakari) to update the build files
when there's a necessary change. I've added a step to CI that checks
whether the workspace-hack crate is up to date, and instructs you to
re-run `script/update-workspace-hack` when it fails.
Finally, to make sure that people can still depend on crates in our
workspace without pulling in all the workspace deps, we use a `[patch]`
section following [hakari's
instructions](https://docs.rs/cargo-hakari/0.9.36/cargo_hakari/patch_directive/index.html)
One possible followup task would be making guppy use our
`rust-toolchain.toml` instead of having to duplicate that list in its
config, I opened an issue for that upstream: guppy-rs/guppy#481.
TODO:
- [x] Fix the extension test failure
- [x] Ensure the dev dependencies aren't being unified by Hakari into
the main dependencies
- [x] Ensure that the remote-server binary continues to not depend on
LibSSL
Release Notes:
- N/A
---------
Co-authored-by: Mikayla <mikayla@zed.dev>
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
This PR completes the process of moving git repository state storage and
scanning logic from the worktree crate to `project::git_store`.
Release Notes:
- N/A
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Conrad <conrad@zed.dev>
Display of git statuses in the git panel, project panel, and tabs
regressed in #27391, causing us to frequently see stale statuses. This
turns out to be because we were not emitting the
`WorktreeUpdatedGitRepositories` event in cases where we should be,
which in turn is because of bumping the `LocalRepositoryEntry`'s
`status_scan_id` too early, so that a later comparison of two
`status_scan_id` values wasn't detecting a change that we're expecting
it to detect.
Release Notes:
- N/A (problematic behavior didn't make it into stable or preview)
This is another in the series of PRs to make the GitStore own all
repository state and enable better concurrency control for git
repository scans.
After this PR, the `RepositoryEntry`s stored in worktree snapshots are
used only as a staging ground for local GitStores to pull from after
git-related events; non-local worktrees don't store them at all,
although this is not reflected in the types. GitTraversal and other
places that need information about repositories get it from the
GitStore. The GitStore also takes over handling of the new
UpdateRepository and RemoveRepository messages. However, repositories
are still discovered and scanned on a per-worktree basis, and we're
still identifying them by the (worktree-specific) project entry ID of
their working directory.
- [x] Remove WorkDirectory from RepositoryEntry
- [x] Remove worktree IDs from repository-related RPC messages
- [x] Handle UpdateRepository and RemoveRepository RPCs from the
GitStore
Release Notes:
- N/A
---------
Co-authored-by: Max <max@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Temporary Workaround For: #27283
This PR can (and should!) be reverted once the underlying inefficiencies
are resolved
Release Notes:
- Files that are 6GB or larger will now not open. This is a temporary
workaround for inefficient handling of large files resulting in
extremely high memory usage, often resulting in system freezing,
requiring a restart of Zed or the entire system.
This is a pure refactoring PR that goes through all the git-related APIs
exposed by the worktree crate and minimizes their use outside that
crate, migrating callers of those APIs to read from the GitStore
instead. This is to prepare for evacuating git repository state from
worktrees and making the GitStore the new source of truth.
Other drive-by changes:
- `project::git` is now `project::git_store`, for consistency with the
other project stores
- the project panel's test module has been split into its own file
Release Notes:
- N/A
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
This PR updates our DB schemas and wire protocol to separate the
synchronization of git statuses and other repository state from the
synchronization of worktrees. This paves the way for moving the code
that executes git status updates out of the `worktree` crate and onto
the new `GitStore`. That end goal is motivated by two (related) points:
- Disentangling git status updates from the worktree's
`BackgroundScanner` will allow us to implement a simpler concurrency
story for those updates, hopefully fixing some known but elusive bugs
(upstream state not updating after push; statuses getting out of sync in
remote projects).
- By moving git repository state to the project-scoped `GitStore`, we
can get rid of the duplication that currently happens when two worktrees
are associated with the same git repository.
Co-authored-by: Max <max@zed.dev>
Release Notes:
- N/A
---------
Co-authored-by: Max <max@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
This PR reworks the `FakeGitRepository` type that we use for testing git
interactions, to make it more realistic. In particular, the `status`
method now derives the Git status from the differences between HEAD, the
index, and the working copy. This way, if you modify a file in the
`FakeFs`, the Git repository's `status` method will reflect that
modification.
Release Notes:
- N/A
---------
Co-authored-by: Junkui Zhang <364772080@qq.com>
This is the core change:
https://github.com/zed-industries/zed/pull/26758/files#diff-044302c0d57147af17e68a0009fee3e8dcdfb4f32c27a915e70cfa80e987f765R1052
TODO:
- [x] Use AsyncFn instead of Fn() -> Future in GPUI spawn methods
- [x] Implement it in the whole app
- [x] Implement it in the debugger
- [x] Glance at the RPC crate, and see if those box future methods can
be switched over. Answer: It can't directly, as you can't make an
AsyncFn* into a trait object. There's ways around that, but they're all
more complex than just keeping the code as is.
- [ ] Fix platform specific code
Release Notes:
- N/A
This PR adds a failing test `test_staging_hunks_with_delayed_fs_event`
and makes it pass
Also skips a queued read for git diff states if another read was
requested (less work)
This still doesn't catch all race conditions, but the PR is getting long
so I'll yield this and start another branch
Release Notes:
- N/A
This was added in #8343 to make it only visible for tests. #9189 then
made it visible regardless of `test-support`, so the definitions became
identical.
Release Notes:
- N/A
I thought it might be just `test_file_status` this time, but it seems to
be all four of the tests that we were previously seeing issues with.
Release Notes:
- N/A
This moves spawning of the git subprocess to the main thread. We're not
yet
sure why, but when we spawn a process using GCD's background queues,
sub-processes like git-credential-manager fail to open windows.
This seems to be fixable either by using the main thread, or by using a
standard background thread,
but for now we use the main thread.
Release Notes:
- Git: Fix git-credential-manager
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Kirill Bulatov <mail4score@gmail.com>
This reduces the number of multibuffer syncs from 100,000 to 20,000.
Before this change each editor individually observed the project, so
literally any project change was amplified by the number of editors you
had open.
Now editors listen to their buffers instead of the project, and other
users of `cx.observe` on the project have been updated to use specific
events to reduce churn.
Follow up to #26237
Release Notes:
- Improved performance of Zed in large repos with lots of file system
events.
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Modified version of #25950. We still use worktree paths, but repo paths
with a status that lie outside the worktree are not excluded; instead,
we relativize them by adding `..`. This makes the list in the git panel
match what you'd get from running `git status` (with the repo's worktree
root as the working directory).
- [x] Implement + test new unrelativization logic
- [x] ~~When collecting repositories, dedup by .git abs path, so
worktrees can share a repo at the project level~~ dedup repos at the
repository selector layer, with repos coming from larger worktrees being
preferred
- [x] Open single-file worktree with diff when activating a path not in
the worktree
Release Notes:
- N/A
Closes#25976
There's a couple states that this covers:
- upon `git init`, no footer is shown at all
- after 1 commit (or when on any parentless commit), the uncommit button
is ~disabled~ hidden
- otherwise commit button is shown
Also updated the button with "meta" tooltip showing human readable
description and git command.
Release Notes:
- N/A
---------
Co-authored-by: Nate Butler <iamnbutler@gmail.com>
Attempt to fix an issue where conflicts from a cherry-pick don't get
cleared out of the git panel after being resolved.
Release Notes:
- Git Beta: Fixed resolution of conflicts from cherry-picks not being
reflected in the git panel
When a worktree is created, we walk up the ancestors of the root path
trying to find a git repository. In particular, if your `$HOME` is a git
repository and you open some subdirectory of `$HOME` that's *not* a git
repository, we end up scanning `$HOME` and everything under it looking
for changed and untracked files, which is often pretty slow. Consistency
here is not very useful and leads to a bad experience.
This PR adds a special case to not consider `$HOME` as a containing git
repository, unless you ask for it by doing the equivalent of `zed ~`.
Release Notes:
- Changed the behavior of git features to not treat `$HOME` as a git
repository unless opened directly
We've had some issues reported with git repositories not getting
detected when they're a strict parent of the worktree root. Add a bit
more logging to understand what's going on here.
Release Notes:
- N/A
Closes https://github.com/zed-industries/zed/issues/25865
Closes https://github.com/zed-industries/zed/pull/25915
In the issue, Zed had caused `.git/lfs/tmp/466102258`-like files to
appear in the directory, which lead to background FS event listener to
handle this as an update, incrementing snapshot's `scan_id`, which lead
to git status rescan, which caused another increment to `status_scan_id`
— incrementing either of the IDs causes the related repo data to be
considered "changed:
41b45eaba7/crates/worktree/src/worktree.rs (L1590-L1605)
hence propagating events to the other parts of the system (e.g. git
blame, which was also active in the issue's case)
```
[2025-03-01T20:01:08+01:00 DEBUG worktree] ignoring event ".git/lfs/tmp/466102258" within unloaded directory
[2025-03-01T20:01:08+01:00 DEBUG worktree] received fs events []
[2025-03-01T20:01:08+01:00 DEBUG worktree] reloading repositories: ["/Users/alex/dev/monorepo/.git"]
[2025-03-01T20:01:08+01:00 DEBUG editor::blame] Status of git repositories updated. Regenerating blame data...
[2025-03-01T20:01:08+01:00 DEBUG editor::blame] Status of git repositories updated. Regenerating blame data...
[2025-03-01T20:01:08+01:00 DEBUG editor::blame] Status of git repositories updated. Regenerating blame data...
```
Due to repo update events sent, another `.git/lfs/tmp/` entry is
created, things start over...
The PR fixes this by ignoring any `.git/lfs/` directory-related FS
events, as needed for the current git status update heuristics.
https://github.com/zed-industries/zed/pull/25915 tried to follow further
and `scan_id` and `status_scan_id` but we do not store all git state in
memory, e.g. head
e0060b92cc/crates/editor/src/editor_tests.rs (L13686)
as
[tests](https://github.com/zed-industries/zed/actions/runs/13631960559/job/38101504549?pr=25915)
show.
Release Notes:
- Improved `.git` scan heuristics
Like the real app, this one infinite loops if you have a diff in an
UnsharedFile.
Release Notes:
- N/A *or* Added/Fixed/Improved ...
---------
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
