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 makes it so the tool selector will stay open when toggling tools
instead of closing after each selection:
https://github.com/user-attachments/assets/eb987785-cfb5-4b07-8d63-510fbd9d9bf1
This involved making a change to `ContextMenu` to allow it to rebuild
its menu items after each confirmation in order for them to reflect
their selected/unselected status. I intend to clean up the `ContextMenu`
API a bit at a later point, but that is out of scope for this PR.
Release Notes:
- N/A
In the process of adding `@mentions` we realized that we do not want to
make a distinction between Files & Directories in the UI, therefore this
PR combines the File & Directory pickers into a unified version
https://github.com/user-attachments/assets/f3bf189c-8b69-4f5f-90ce-0b83b12dbca3
(Ignore the `@mentions`, they are broken also on main)
Release Notes:
- N/A
When the model reads file, we'll track the version it read, and let it
know if the user makes edits to the buffer. This helps prevent edit
failures because it'll know to re-read the file before.
Release Notes:
- N/A
Similar to how tasks are fetched via LSP, also queries for document's
code lens and filters the ones with the commands, supported in server
capabilities.
Whatever's left and applicable to the range given, is added to the
actions menu:
This way, Zed can get more actions to run, albeit neither r-a nor vtsls
seem to provide anything by default.
Currently, there are no plans to render code lens the way as in VSCode,
it's just the extra actions that are show in the menu.
------------------
As part of the attempts to use rust-analyzer LSP data about the
runnables, I've explored a way to get this data via standard LSP.
When particular experimental client capabilities are enabled (similar to
how clangd does this now), r-a starts to send back code lens with the
data needed to run a cargo command:
```
{"jsonrpc":"2.0","id":48,"result":{"range":{"start":{"line":0,"character":0},"end":{"line":98,"character":0}},"command":{"title":"▶︎ Run Tests","command":"rust-analyzer.runSingle","arguments":[{"label":"test-mod tests::ecparser","location":{"targetUri":"file:///Users/someonetoignore/work/ec4rs/src/tests/ecparser.rs","targetRange":{"start":{"line":0,"character":0},"end":{"line":98,"character":0}},"targetSelectionRange":{"start":{"line":0,"character":0},"end":{"line":98,"character":0}}},"kind":"cargo","args":{"environment":{"RUSTC_TOOLCHAIN":"/Users/someonetoignore/.rustup/toolchains/1.85-aarch64-apple-darwin"},"cwd":"/Users/someonetoignore/work/ec4rs","overrideCargo":null,"workspaceRoot":"/Users/someonetoignore/work/ec4rs","cargoArgs":["test","--package","ec4rs","--lib"],"executableArgs":["tests::ecparser","--show-output"]}}]}}}
```
This data is passed as is to VSCode task processor, registered in
60cd01864a/editors/code/src/main.ts (L195)
where it gets eventually executed as a VSCode's task, all handled by the
r-a's extension code.
rust-analyzer does not declare server capabilities for such tasks, and
has no `workspace/executeCommand` handle, and Zed needs an interactive
terminal output during the test runs, so we cannot ask rust-analyzer
more than these descriptions.
Given that Zed needs experimental capabilities set to get these lens:
60cd01864a/editors/code/src/client.ts (L318-L327)
and that the lens may contain other odd tasks (e.g. docs opening or
references lookup), a protocol extension to get runnables looks more
preferred than lens:
https://rust-analyzer.github.io/book/contributing/lsp-extensions.html#runnables
This PR does not include any work on this direction, limiting to the
general code lens support.
As a proof of concept, it's possible to get the lens and even attempt to
run it, to no avail:
Release Notes:
- Used `textDocument/codeLens` data in the actions menu when applicable
When the user attached context in the thread, the editor model request
would fail because its tool use wouldn't be removed properly leading to
an API error.
Also, after an edit, we'd keep the old file snapshot in the context.
This would make the model think that the edits didn't apply and make it
go in a loop.
Release Notes:
- N/A
Release Notes:
- Multibuffers now use less vertical space for excerpt boundaries.
Additionally the expand up/down arrows are hidden at the start and end
of the buffers
---------
Co-authored-by: Nate Butler <iamnbutler@gmail.com>
Co-authored-by: Zed AI <claude-3.5-sonnet@zed.dev>
This PR updates the `Thread::is_streaming` method so that it includes
tool use in the "streaming" state.
This will prevent the streaming indicator from disappearing when we're
doing tool use.
Release Notes:
- N/A
This PR cleans up some color & elevation misc.
### Don't allow deriving Color from Hsla
The point of the [ui::Color] enum is to encourage consistent color
usage, and the the Color::Custom case is really only meant for cases
where we have no other choice.
`impl From<Hsla> for Color` encourages blindly passing colors into
`Color::Custom` – with this in place we might as well remove the entire
`Color` enum.
The usages that were updated due to this removal were for colors that
already exist in the Color enum, making it even more clear that it
didn't make sense to have this.
### `ElevationIndex` -> `Elevation`
This name would make more sense if we had an `Elevation` in the first
place. The new name is more clear.
#### `Button::elevation`
As part of this change I also updated button's `layer` method to
`elevation`, since it takes an elevation. This method still has the
following issue:
You want to use `Button::elevation` when it's default colors are
invisible on the layer you are rendering the button on. However, current
this method uses the elevation's `bg` color, rather than it's
`on_elevation_bg`.
Ideally when you use `Button::elevation` you want to pass the elevation
you are _on_, not choosing one that will show up the elevation you are
on.
This change will be in a separate PR, as it likely will have widespread
visual impact across the app.
Release Notes:
- N/A
This should be less eager in terms of invoking tools. But we should keep
iterating on it as we add more tools.
Also, this disables the Lua interpreter by default (it can still be
enabled manually from the tools icon).
Release Notes:
- N/A
---------
Co-authored-by: Richard Feldman <oss@rtfeldman.com>
Exposes a new "edit files" tool that the model can use to apply
modifications to files in the project. The main model provides
instructions and the tool uses a separate "editor" model (Claude 3.5 by
default) to generate search/replace blocks like Aider does:
````markdown
mathweb/flask/app.py
```python
<<<<<<< SEARCH
from flask import Flask
=======
import math
from flask import Flask
>>>>>>> REPLACE
```
````
The search/replace blocks are parsed and applied as they stream in. If a
block fails to parse, the tool will apply the other edits and report an
error pointing to the part of the input where it occurred. This should
allow the model to fix it.
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
https://github.com/user-attachments/assets/a486ff2a-4aa1-4c0d-be6c-1dea2a8d60c8
- [x] Track buffer changes in `ScriptingSession`
- [x] Show edited files in thread
Reviewing diffs and displaying line counts will be part of an upcoming
PR.
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
This PR decouples the scripting tool from the `Tool` trait while still
allowing it to be used as a tool from the model's perspective.
This will allow us to evolve the scripting tool as more of a first-class
citizen while still retaining the ability to have the model call it as a
regular tool.
Release Notes:
- N/A
We decided to expose scripting as tools again. We are aware of the UX
downsides of doing so, but we want to focus on getting it working well
first, and the model seems to make better use of it as an actual tool.
In the future, the tools API might support streaming. If it doesn't and
we need to ship, we can consider reverting this.
Release Notes:
- N/A
- Truncate branch names based on the width of the picker
- Use a footer for "Create branch" instead of a picker entry
Still to do:
- [x] Select the footer button when no matches and run the create logic
on `enter`
- [x] Make it possible to quickly select the footer button from the
keyboard when there are matches
Release Notes:
- Git Beta: Removed limitation that made it impossible to create a
branch from the branch picker when it too closely resembled an existing
branch name
This PR makes refactors the scripting functionality to be a first-class
concept of the assistant instead of a generic tool, which will allow us
to build a more customized experience.
- The tool prompt has been slightly tweaked and is now included as a
system message in all conversations. I'm getting decent results, but now
that it isn't in the tools framework, it will probably require more
refining.
- The model will now include an `<eval ...>` tag at the end of the
message with the script. We parse this tag incrementally as it streams
in so that we can indicate that we are generating a script before we see
the closing `</eval>` tag. Later, this will help us interpret the script
as it arrives also.
- Threads now hold a `ScriptSession` entity which manages the state of
all scripts (from parsing to exited) in a centralized way, and will
later collect all script operations so they can be displayed in the UI.
- `script_tool` has been renamed to `assistant_scripting`
- Script source now opens in a regular read-only buffer
Note: We still need to handle persistence properly
Release Notes:
- N/A
---------
Co-authored-by: Marshall Bowers <git@maxdeviant.com>
This PR factors out a new `Thread::all_tools_finished` method to
encapsulate some of the boilerplate in the `ThreadEvent::ToolFinished`
event handler.
This should make this event handler easier to replicate for the eval
use-case.
Release Notes:
- N/A
