Closes#14146. This is just for the time being before a networking
refactoring to split reads and writes on the ROUTER/DEALER ZeroMQ
sockets. Some kernels have not been responding with `kernel_info_reply`,
which ends up hanging our shell socket.
Release Notes:
- N/A
Release notes for the REPL feature will be part of its official launch.
Sets up the `cmd-enter` keybinding for the jupyter repl to only apply
when enabled.
Release Notes:
- N/A
---------
Co-authored-by: Kirill <kirill@zed.dev>
Initial runtimes UI panel. The main draw here is that all message
subscription occurs with two background tasks that run for the life of
the kernel. Follow on to #12062
* [x] Disable previous cmd-enter behavior only if runtimes are enabled
in settings
* [x] Only show the runtimes panel if it is enabled via settings
* [x] Create clean UI for the current sessions
### Running Kernels UI
<img width="205" alt="image"
src="https://github.com/zed-industries/zed/assets/836375/814ae79b-0807-4e23-bc95-77ce64f9d732">
* [x] List running kernels
* [x] Implement shutdown
* [x] Delete connection file on `drop` of `RunningKernel`
* [x] Implement interrupt
#### Project-specific Kernel Settings
- [x] Modify JupyterSettings to include a `kernel_selections` field
(`HashMap<String, String>`).
- [x] Implement saving and loading of kernel selections to/from
`.zed/settings.json` (by default, rather than global settings?)
#### Kernel Selection Persistence
- [x] Save the selected kernel for each language when the user makes a
choice.
- [x] Load these selections when the RuntimePanel is initialized.
#### Use Selected Kernels
- [x] Modify kernel launch to use the selected kernel for the detected
language.
- [x] Fallback to default behavior if no selection is made.
### Empty states
- [x] Create helpful UI for when the user has 0 kernels they can launch
and/or 0 kernels running
<img width="694" alt="image"
src="https://github.com/zed-industries/zed/assets/836375/d6a75939-e4e4-40fb-80fe-014da041cc3c">
## Future work
### Kernel Discovery
- Improve the kernel discovery process to handle various installation
methods (system, virtualenv, poetry, etc.).
- Create a way to refresh the available kernels on demand
### Documentation:
- Update documentation to explain how users can configure kernels for
their projects.
- Provide examples of .zed/settings.json configurations for kernel
selection.
### Kernel Selection UI
- Implement a new section in the RuntimePanel to display available
kernels.
- Group on the language name from the kernel specification
- Create a dropdown for each language group to select the default
kernel.
Release Notes:
- N/A
---------
Co-authored-by: Kirill <kirill@zed.dev>
Run any Jupyter kernel in Zed on any buffer (editor):
<img width="1074" alt="image"
src="https://github.com/zed-industries/zed/assets/836375/eac8ed69-d02b-4d46-b379-6186d8f59470">
## TODO
### Lifecycle
* [x] Launch kernels on demand
* [x] Wait for kernel to be started
* [x] Request Kernel info on start
* [x] Show in progress indicator
* [ ] Allow picking kernel (it defaults to first matching language name)
* [ ] Menu for interrupting and shutting down the kernel
* [ ] Drop running kernels once editor is dropped
### Media Outputs
* [x] Render text and tracebacks with ANSI color handling
* [x] Render markdown as text
* [x] Render PNG and JPEG images using an explicit height based on
line-height
* ~~Render SVG~~ -- not happening for this PR due to lack of text in SVG
support
* [ ] Process `update_display_data` message and related `display_id`
* [x] Process `page` data from payloads as outputs
* [ ] Render markdown as, well, rendered markdown -- Note: unsure if we
can get line heights here
### Document
* [x] Select code and run
* [x] Run current line
* [x] Clear previous overlapping runs
* [ ] Support running markdown code blocks
* [ ] Action to export session as notebook or output files
* [ ] Action to clear all outputs
* [ ] Delete outputs when lines are deleted
## Other missing features
The following is a list of missing functionality or expectations that
are out of scope for this PR.
### Python Environments
Detecting python environments should probably be done in a separate PR
in tandem with how they're used with LSP. Users likely want to pick an
environment for their project, whether a virtualenv, conda env, pyenv,
poetry backed virtualenv, or the system. Related issues:
* https://github.com/zed-industries/zed/issues/7646
* https://github.com/zed-industries/zed/issues/7808
* https://github.com/zed-industries/zed/issues/7296
### LSP Integration
* Submit `complete_request` messages for completions to interleave
interactive variables with LSP
* LSP for IPython semantics (`%%timeit`, `!ls`, `get_ipython`, etc.)
## Future release notes
- Run code in any editor, whether it's a script or a markdown document
Release Notes:
- N/A
This adds the ability to set the keep alive as an integer, including
`-1` for staying alive indefinitely until a new model is loaded or
Ollama exits. I've also set the default to `-1` so that models stay
ready to go for Zed to use.
Release Notes:
- N/A
Closes#4424.
A few design decisions that may need some rethinking or later PRs:
* Other providers have a check for authentication. I use this
opportunity to fetch the models which doubles as a way of finding out if
the Ollama server is running.
* Ollama has _no_ API for getting the max tokens per model
* Ollama has _no_ API for getting the current token count
https://github.com/ollama/ollama/issues/1716
* Ollama does allow setting the `num_ctx` so I've defaulted this to
4096. It can be overridden in settings.
* Ollama models will be "slow" to start inference because they're
loading the model into memory. It's faster after that. There's no UI
affordance to show that the model is being loaded.
Release Notes:
- Added an Ollama Provider for the assistant. If you have
[Ollama](https://ollama.com/) running locally on your machine, you can
enable it in your settings under:
```jsonc
"assistant": {
"version": "1",
"provider": {
"name": "ollama",
// Recommended setting to allow for model startup
"low_speed_timeout_in_seconds": 30,
}
}
```
Chat like usual
<img width="1840" alt="image"
src="https://github.com/zed-industries/zed/assets/836375/4e0af266-4c4f-4d9e-9d74-1a91f76a12fe">
Interact with any model from the [Ollama
Library](https://ollama.com/library)
<img width="587" alt="image"
src="https://github.com/zed-industries/zed/assets/836375/87433ac6-bf87-4a99-89e1-96a93bf8de8a">
Open up the terminal to download new models via `ollama pull`:
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 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
Fixes#11422 by accepting just the start of the line.
Release Notes:
- N/A
---------
Co-authored-by: max <max@zed.dev>
Co-authored-by: jacob <jacob@supermaven.com>
Adds a supermaven provider for completions. There are various other
refactors amidst this branch, primarily to make copilot no longer a
dependency of project as well as show LSP Logs for global LSPs like
copilot properly.
This feature is not enabled by default. We're going to seek to refine it
in the coming weeks.
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Max <max@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Switches Assistant2 to always provide functions. It's up to the model to
choose to use them. At a later point, the `Submit::Codebase` should
change the `tool_choice` to `query_codebase` rather than `auto`. For
now, I think this will improve the experience for folks testing.
Release Notes:
- N/A
Creates a `ToolView` trait to allow interactivity. This brings expanding
and collapsing to the excerpts from project index searches.
Release Notes:
- N/A
---------
Co-authored-by: Nathan <nathan@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
This is a crate only addition of a new version of the AssistantPanel.
We'll be putting this behind a feature flag while we iron out the new
experience.
Release Notes:
- N/A
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Conrad Irwin <conrad@zed.dev>
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
Co-authored-by: Antonio Scandurra <antonio@zed.dev>
Co-authored-by: Nate Butler <nate@zed.dev>
Co-authored-by: Nate Butler <iamnbutler@gmail.com>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Max <max@zed.dev>
Updates `collab` to accept a `seed.json` file that allows you to
override the defaults. Updated the `README` in collab to just have
directions inside instead of redirecting the developer to the website.
Release Notes:
- N/A
Co-authored-by: Max <max@zed.dev>
This fixes when the workspace is not actually available for a
`.read(cx)`.
Release Notes:
- Fix a panic when quoting a selection before the assistant panel has
been started
Co-authored-by: Conrad Irwin <conrad@zed.dev>
This introduces semantic indexing in Zed based on chunking text from
files in the developer's workspace and creating vector embeddings using
an embedding model. As part of this, we've created an embeddings
provider trait that allows us to work with OpenAI, a local Ollama model,
or a Zed hosted embedding.
The semantic index is built by breaking down text for known
(programming) languages into manageable chunks that are smaller than the
max token size. Each chunk is then fed to a language model to create a
high dimensional vector which is then normalized to a unit vector to
allow fast comparison with other vectors with a simple dot product.
Alongside the vector, we store the path of the file and the range within
the document where the vector was sourced from.
Zed will soon grok contextual similarity across different text snippets,
allowing for natural language search beyond keyword matching. This is
being put together both for human-based search as well as providing
results to Large Language Models to allow them to refine how they help
developers.
Remaining todo:
* [x] Change `provider` to `model` within the zed hosted embeddings
database (as its currently a combo of the provider and the model in one
name)
Release Notes:
- N/A
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Conrad Irwin <conrad@zed.dev>
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
Co-authored-by: Antonio <antonio@zed.dev>
Absent some ability to toggle between viewing and editing a file, I
think it would be best to get a fix out quick for people to edit SVGs as
text files.
Release Notes:
- Fixed editing of SVG images by disabling it from the image viewer
([#10403](https://github.com/zed-industries/zed/issues/10403)).
While working towards fixes for the image viewer, @mikayla-maki and I
discovered that we didn't have `object-fit: scale-down` implemented.
This doesn't _fully_ solve the image issues as there is some issue where
only the bounds width is updating on layout change that I haven't fully
chased down.
Co-Authored-By: @mikayla-maki
Release Notes:
- N/A
---------
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
Introducing the Active File Context portion of #9705. When someone is in
the assistant panel it now includes the active file as a system message
on send while showing them a nice little display in the lower right:
For this iteration, I'd love to see the following before we land this:
* [x] Toggle-able context - user should be able to disable sending this
context
* [x] Show nothing if there is no context coming in
* [x] Update token count as we change items
* [x] Listen for a more finely scoped event for when the active item
changes
* [x] Create a global for pulling a file icon based on a path. Zed's
main way to do this is nested within project panel's `FileAssociation`s.
* [x] Get the code fence name for a Language for the system prompt
* [x] Update the token count when the buffer content changes
I'm seeing this PR as the foundation for providing other kinds of
context -- diagnostic summaries, failing tests, additional files, etc.
Release Notes:
- Added file context to assistant chat panel
([#9705](https://github.com/zed-industries/zed/issues/9705)).
<img width="1558" alt="image"
src="https://github.com/zed-industries/zed/assets/836375/86eb7e50-3e28-4754-9c3f-895be588616d">
---------
Co-authored-by: Conrad Irwin <conrad@zed.dev>
Co-authored-by: Nathan <nathan@zed.dev>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
This is the beginning of setting up a flexible way to open items beyond
the text buffer -- think notebooks, images, GeoJSON, etc. The primary
requirement to allow opening an arbitrary file is `try_open` on the
`project::Item` trait. Now we can make new `Item`s for other types with
their own ways to render.
Under the hood, `register_project_item` uses this new opening scheme. It
supports a dynamic array of opener functions, that will handle specific
item types. By default, a `Buffer` should be able to be able to open any
file that another opener did not.
A key detail here is that the order of registration matters. The last
item has primacy. Here's an example:
```rust
workspace::register_project_item::<Editor>(cx);
workspace::register_project_item::<Notebook>(cx);
workspace::register_project_item::<ImageViewer>(cx);
```
When a project item (file) is attempted to be opened, it's first tried
with the `ImageViewer`, followed by the `Notebook`, then the `Editor`.
The tests are set up in a way that should make it _hopefully_ easy to
learn how to write a new opener. First to go after should probably be
image files.
Release Notes:
N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
Kyle Kelley
