This PR adds support to Assistant 2 for using tools provided by context
servers.
As part of this I introduced a new `ThreadStore`.
Release Notes:
- N/A
---------
Co-authored-by: Cole <cole@zed.dev>
This PR restructures the storage of the tool uses and results in
`assistant2` so that they don't live on the individual messages.
It also introduces a `LanguageModelToolUseId` newtype for better type
safety.
Release Notes:
- N/A
This PR adds rudimentary support for using tools to `assistant2`. There
are currently no visual affordances for tool use.
This is gated behind the `assistant-tool-use` feature flag.
<img width="1079" alt="Screenshot 2024-11-25 at 7 21 31 PM"
src="https://github.com/user-attachments/assets/64d6ca29-c592-4474-8e9d-c344f855bc63">
Release Notes:
- N/A
This PR makes it so previous messages in the thread are included when
constructing the completion request, instead of only sending up the most
recent user message.
Release Notes:
- N/A
This PR breaks ground on a new `assistant2` crate.
In order to see this new version of the assistant, both of the following
must be true:
1. The `assistant2` feature flag is enabled for your user
- It is **not** currently enabled for all staff.
2. You are running a development build of Zed
The intent here is to enable the folks working on `assistant2` to
incrementally land work onto `main` without breaking use of the current
Assistant for anyone.
<img width="1136" alt="Screenshot 2024-11-23 at 10 46 08 AM"
src="https://github.com/user-attachments/assets/5723a13f-5be1-4486-9460-ead7329ba78e">
Release Notes:
- N/A
Release Notes:
- Improved expand excerpt indicators to allow unidirectional expansion.
Also added the `editor::ExpandExcerptsUp` and
`editor::ExpandExcerptsDown` actions, which can both take a `lines`
parameter. Also added a `expand_excerpt_lines` setting which controls
the default number of lines that the indicators and actions use.
---------
Co-authored-by: conrad <conrad@zed.dev>
This pull request replaces the static `⋯` character we used to insert
when folding a range with a custom render function that return an
`AnyElement`. We plan to use this in the assistant, but for now this
should be behavior-preserving.
Release Notes:
- N/A
---------
Co-authored-by: Nathan <nathan@zed.dev>
Co-authored-by: Conrad <conrad@zed.dev>
This PR adds some ergonomic improvements when working with GPUI
`Global`s.
Two new traits have been added—`ReadGlobal` and `UpdateGlobal`—that
provide associated functions on any type that implements `Global` for
accessing and updating the global without needing to call the methods on
the `cx` directly (which generally involves qualifying the type).
I looked into adding `ObserveGlobal` as well, but this seems a bit
trickier to implement as the signatures of `cx.observe_global` vary
slightly between the different contexts.
Release Notes:
- N/A
Additionally, the internal `ToolView` trait used by the registry is now
called `InternalToolView`.
This should make it a bit easier to understand that the `ToolView` is
intended for a `gpui::View` (implementing `Render`). It does still feel
like more could be merged here but I think the built tools are now a bit
clearer.
Release Notes:
- N/A
Improves the descriptions for some of the tools. I wish we had metrics
to back up changes in how the model responds to tool schema changes so
anecdotally I'm just going to say this _seems_ improved.
Release Notes:
- N/A
This PR updates the saved conversation picker to use a list of
conversations retrieved from disk instead of the static placeholder
values.
Release Notes:
- N/A
This shows "Researching..." as placeholder text as early as possible so
that the user can see the model is working on reading/researching/etc.
This also adds on an `Option<Value>` to the `render_running` function so
that tools can hopefully render based on partially completed JSON (still
to come).
Release Notes:
- N/A
This PR updates the new assistant with a button to start a new
conversation.
Clicking on it will reset the chat and put it into a fresh state.
The current conversation will be serialized and written to
`~/.config/zed/conversations`.
Release Notes:
- N/A
This PR adds the general structure for conversation history to the new
assistant.
Right now we have a placeholder button in the assistant panel that will
toggle a picker containing some placeholder saved conversations.
Release Notes:
- N/A
This PR restructures the way that tools and attachments add information
about the current project to a conversation with the assistant. Rather
than each tool call or attachment generating a new tool or system
message containing information about the project, they can all
collectively mutate a new type called a `ProjectContext`, which stores
all of the project data that should be sent to the assistant. That data
is then formatted in a single place, and passed to the assistant in one
system message.
This prevents multiple tools/attachments from including redundant
context.
Release Notes:
- N/A
---------
Co-authored-by: Kyle <kylek@zed.dev>
This PR adjusts the names of the negated style methods by moving the
`neg_` to after the property name instead of before.
This will help keep related style methods grouped together in
completions.
It also makes it a bit clearer that the negation applies to the value.
### Before
```rs
div()
.neg_mx_1()
.neg_mt_2()
```
### After
```rs
div()
.mx_neg_1()
.mt_neg_2()
```
Release Notes:
- N/A
This PR makes the `border` methods require an explicit width instead of
defaulting to 1px.
This breaks convention with Tailwind, but it makes GPUI more consistent
with itself. We already have an edge case where the parameterized method
had to be named `border_width`, since `border` was taken up by an alias
for the 1px variant.
### Before
```rs
div()
.border()
.border_t()
.border_r()
.border_b()
.border_l()
.border_width(px(7.))
```
### After
```rs
div()
.border_1()
.border_t_1()
.border_r_1()
.border_b_1()
.border_l_1()
.border(px(7.))
```
Release Notes:
- N/A
The CI was green at the time I've merged Rust 1.78, but a change that
violated clippy::empty_doc has slipped through into main in the
meantime. Mea culpa, I should've reran the CI.
Release Notes:
- N/A
Notable things I've had to fix due to 1.78:
- Better detection of unused items
- New clippy lint (`assigning_clones`) that points out places where assignment operations with clone rhs could be replaced with more performant `clone_into`
Release Notes:
- N/A
This fixes a tricky intermittent issue I was seeing, where failed to
chunk certain files correctly because of the way we reuse Tree-sitter
`Parser` instances across parses.
I've also accounted for leading comments in chunk boundaries, so that
items are grouped with their leading comments whenever possible when
chunking.
Finally, we've changed the `debug project index` action so that it opens
a simple debug view in a pane, instead of printing paths to the console.
This lets you click into a path and see how it was chunked.
Release Notes:
- N/A
---------
Co-authored-by: Marshall <marshall@zed.dev>
