Closes#30678
This is caused by `TerminalOutput::full_text` triming trailing newline
when creating the "REPL Output" buffer.
Release Notes:
- fix: Preserve trailing newline in `TerminalOutput::full_text`
The current problem is that if I specify model parameters, like
`max_tokens`, in `settings.json` for an Ollama model, they do not
override the values coming from the Ollama API. Instead, the parameters
from the API are used. For example, in the settings below, even though I
have overridden `max_tokens`, Zed will still use the API's default
`context_length` of 4k.
```
"language_models": {
"ollama": {
"available_models": [
{
"name": "qwen3-coder:latest",
"display_name": "Qwen 3 Coder",
"max_tokens": 64000,
"supports_tools": true,
"keep_alive": "15m",
"supports_thinking": false,
"supports_images": false
}
]
}
},
```
Release Notes:
- Fixed an issue where Ollama model parameters were not being correctly
overridden by user settings.
Closes https://github.com/zed-industries/zed/issues/37882
Previously, every piece of text in the agent panel was controlled by
`agent_font_size`. Although it is nice to only have one setting to tweak
that, it could be a bit misleading particularly because we use
monospaced and sans-serif fonts for different elements in the panel. Any
editor/textarea in the panel, whehter it is the main message editor or
the previous message editor, uses the buffer font. Therefore, I think it
is reasonable to expect that tweaking `buffer_font_size` would also
change the agent panel's usage of buffer fonts.
With this change, regular buffers and the agent panel's message editor
will always have the same size.
Release Notes:
- agent: Made the agent panel's textarea font size follow the font size
of regular buffers. They're now both controlled by the
`buffer_font_size` setting.
Address an issue where, in Vim mode, clicking past the end of a line
after selecting the entire line would place the cursor on the newline
character instead of the last character of the line, which is
inconsistent with Vim's normal mode expectations.
I believe the root cause was that the cursor’s position was updated to
the end of the line before the mode switch from Visual to Normal, at
which point `DisplayMap.clip_at_line_ends` was still set to `false`. As
a result, the cursor could end up in an invalid position for Normal
mode. The fix ensures that when switching between these two modes, and
if the selection is empty, the selection point is properly clipped,
preventing the cursor from being placed past the end of the line.
Related #38049
Release Notes:
- Fixed issue in Vim mode where switching from any mode to normal mode
could end up with the cursor in the newline character
---------
Co-authored-by: Conrad Irwin <conrad.irwin@gmail.com>
Nowadays, people don't expect @-mentioning a directory to include the
contents of all files within it. Doing so makes it very likely to
consume an undesirable amount of tokens.
By default, we'll now only include the path of the directory and let the
model decide how much to read via tools. We'll still include the
contents if no tools are available (e.g. "Minimal" profile is selected).
Release Notes:
- Agent Panel: Do not include the content of @-mentioned directories
when tools are available
The edit prediction debug tools has been renamed to zeta2 inspector
because it's now zeta specific. It will now always display the last
prediction request context, prompt, and model response.
Release Notes:
- N/A
---------
Co-authored-by: Bennet <bennet@zed.dev>
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
Closes#38651
`git_panel.set_amend_pending(false, cx);` was being called before
`git_panel.commit_changes(...)` which was causing the commit buffer to
be cleared/reset before actually sending the commit request to git.
Introduced by #35268 which added clear buffer functionality to the
`set_amend_pending` function.
Release Notes:
- Fix git amend on panel sending "Update ..." instead of the original
commit message
- FIx git amend button not working
Release Notes:
- Added "show_summary" & "show_command" settings to the initial
tasks.json file.
This makes the initial task template match the docs here:
https://zed.dev/docs/tasks
Solves an issue where Google APIs refuse all requests with file-based
tool calls attached.
This seems to get triggered in the case where:
- copy_path + another file-based tool call is enabled
- default terminal is `/bin/bash` or something similar
It is unclear why this is happening, but removing the terminal commands
in those tool calls seems to have solved the issue.
Closes#37180 and #37414
Release Notes:
- agent: Fix Gemini refusing requests with certain profiles/systems.
## Problem
Zed was crashing with a UTF-8 character boundary error when rendering
text containing multi-byte characters (like emojis or CJK characters):
```
Thread "main" panicked with "byte index 49 is not a char boundary; it is inside '…' (bytes 48..51)"
```
## Root Cause Analysis
The PR reviewer correctly identified that the issue was not in the
DirectWrite boundary handling, but rather in the text run length
calculation in the text system. When text runs are split across lines in
`text_system.rs:426`, the calculation:
```rust
let run_len_within_line = cmp::min(line_end, run_start + run.len) - run_start;
```
This could result in `run_len_within_line` values that don't respect
UTF-8 character boundaries, especially when multi-byte characters (like
'…' which is 3 bytes) get split across lines. The resulting `FontRun`
objects would have lengths that don't align with character boundaries,
causing the panic when DirectWrite tries to slice the string.
## Solution
Fixed the issue by adding UTF-8 character boundary validation in the
text system where run lengths are calculated. The fix ensures that when
text runs are split across lines, the split always occurs at valid UTF-8
character boundaries:
```rust
// Ensure the run length respects UTF-8 character boundaries
if run_len_within_line > 0 {
let text_slice = &line_text[run_start - line_start..];
if run_len_within_line < text_slice.len() && !text_slice.is_char_boundary(run_len_within_line) {
// Find the previous character boundary using efficient bit-level checking
// UTF-8 characters are at most 4 bytes, so we only need to check up to 3 bytes back
let lower_bound = run_len_within_line.saturating_sub(3);
let search_range = &text_slice.as_bytes()[lower_bound..=run_len_within_line];
// SAFETY: A valid character boundary must exist in this range because:
// 1. run_len_within_line is a valid position in the string slice
// 2. UTF-8 characters are at most 4 bytes, so some boundary exists in [run_len_within_line-3..=run_len_within_line]
let pos_from_lower = unsafe {
search_range
.iter()
.rposition(|&b| (b as i8) >= -0x40)
.unwrap_unchecked()
};
run_len_within_line = lower_bound + pos_from_lower;
}
}
```
## Testing
- ✅ Builds successfully on all platforms
- ✅ Eliminates UTF-8 character boundary panics
- ✅ Maintains existing functionality for all text types
- ✅ Handles edge cases like very long multi-byte characters
## Benefits
1. **Root cause fix**: Addresses the issue at the source rather than
treating symptoms
2. **Performance optimal**: Uses the same efficient algorithm as the
standard library
3. **Minimal changes**: Only modifies the specific problematic code path
4. **Future compatible**: Can be easily replaced with
`str::floor_char_boundary()` when stabilized
## Alternative Approaches Considered
1. **DirectWrite boundary fixing**: Initially tried to fix in
DirectWrite, but this was treating symptoms rather than the root cause
2. **Helper function approach**: Considered extracting to a helper
function, but inlined implementation is more appropriate for this
specific use case
3. **Standard library methods**: `floor_char_boundary()` is not yet
stable, so implemented equivalent logic
The chosen approach provides the best balance of performance, safety,
and code maintainability.
---
Release Notes:
- N/A
Closes
https://github.com/zed-industries/zed/issues/6730#issuecomment-3320933701
That way if multiple servers are running while reporting the same
results we prevent opening multi buffers for single entries.
Release Notes:
- N/A *or* Added/Fixed/Improved ...
Add tests on task commands, to ensure things like
https://github.com/zed-industries/zed/issues/38343 do not come so easily
unnoticed and to provide a base to create more tests in the future, if
needed.
Release Notes:
- N/A
---------
Co-authored-by: Lukas Wirth <lukas@zed.dev>
Closed#33759Closed#38166
### Summary
This PR fixes the way `go test` commands are generated for **testify
suite test methods**.
Previously, only the method name was included in the `-run` flag, which
caused Go’s test runner to fail to find suite test cases.
---
### Problem
https://github.com/user-attachments/assets/e6f80a77-bcf3-457c-8bfb-a7286d44ff71
1. **Incorrect command** was generated for suite tests:
```bash
go test -run TestSomething_Success
```
This results in:
```
testing: warning: no tests to run
```
2. The correct format requires the **suite name + method name**:
```bash
go test -run ^TestFooSuite$/TestSomething_Success$
```
Without the suite prefix (`TestFooSuite`), Go cannot locate test methods
defined on a suite struct.
---
### Changes Made
* **Updated `runnables.scm`**:
* Added a new query rule for suite methods (`.*Suite` receiver types).
* Ensures only methods on suite structs (e.g., `FooSuite`) are matched.
* Tagged these with `go-testify-suite` in addition to `go-test`.
* **Extended task template generation**:
* Introduced `GO_SUITE_NAME_TASK_VARIABLE` to capture the suite name.
* Create a `TaskTemplate` for the testify suite.
* **Improved labeling**:
* Labels now show the full path (`go test ./pkg -v -run
TestFooSuite/TestSomething_Success`) for clarity.
* **Added a test** `test_testify_suite_detection`:
* Covered testify suite cases to ensure correct detection and command
generation.
---
### Impact
https://github.com/user-attachments/assets/ef509183-534a-4aa4-9dc7-01402ac32260
* **Before**: Running a suite test method produced “no tests to run.”
* **After**: Suite test methods are runnable individually with the
correct `-run` command, and full suites can still be executed as before.
### Release Notes
* Fixed generation of `go test` commands for **testify suite test
methods**.
Suite methods now include both the suite name and the method name in the
`-run` flag (e.g., `^TestFooSuite$/TestSomething_Success$`), ensuring
they are properly detected and runnable individually.
Buffer search initiates a new search every time a key is pressed in the
buffer search bar. This would cancel the task associated with any
pending searches. Whenever one of these searches was canceled Zed would
log `[search]: oneshot canceled`. This log would trigger almost on every
keypress when typing moderately fast. This PR silences these logs by not
treating canceled searches as errors.
Release Notes:
- N/A
I noticed that after we paste in line mode, the cursor position is
positioned at the beginning of the next logical line which is somewhat
undesirable since then inserting/appending will position the cursor
after the selection. This does not match helix behaviour which we should
further investigate.
Follow-up to https://github.com/zed-industries/zed/pull/38663
Release Notes:
- N/A
iTerm's editor configuration dialog allows you to set your editor to
`zed \1:\2`, but not (as far as I know) to leave off the : when there's
no line number
This fixes clicking on bare filenames in iTerm for me.
Release Notes:
- Fixed line number parsing so that `zed filename.rs:` will now act as
though you did `zed filename.rs`
Creates a new `EditPredictionProvider` for zeta2, that requests
completions from a new cloud endpoint including context from the new
`edit_prediction_context` crate. This is not ready for use, but it
allows us to iterate.
Release Notes:
- N/A
---------
Co-authored-by: Michael Sloan <michael@zed.dev>
Co-authored-by: Bennet <bennet@zed.dev>
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
This took me quite a while to find out when I developed my first
language extension. I had non-matching entries in the `languages` array
and in the name field of `config.toml`, and it was especially tricky
because the zed extension would start up, but not the language server. I
sure which this had been in the docs, so I am contributing it now!
In particular,
* if the selection ends at the beginning of the next line, and the
current line under the cursor is empty, we paste at the selection's end.
* if however the current line under the cursor is empty, we need to move
to the beginning of the next line to avoid pasting above the end of
current selection
In addition, in line mode, we always move the cursor to the end of the
inserted text. Otherwise, while it looks fine visually,
inserting/appending ends up in the next logical line which is not
desirable.
Release Notes:
- N/A
This PR updates the Gemini provider to treat a
`prompt_feedback.block_reason` as a refusal, as Gemini does not seem to
return a `stop_reason` to use in this case.
<img width="639" height="162" alt="Screenshot 2025-09-22 at 4 23 15 PM"
src="https://github.com/user-attachments/assets/7a86d67e-06c1-49ea-b58f-fa80666f0f8c"
/>
Previously this would just result in no feedback to the user.
Release Notes:
- Added an error message when a Gemini response contains a
`block_reason`.
This PR fixes a regression where settings profiles were no longer
ordered in the same order that the user defined in their settings.
Release Notes:
- N/A
Closes#38495
Cause:
- When the Find input is focused, CopyPath/CopyRelativePath were handled
by the editor and stopped during the bubble phase, preventing
BufferSearchBar from relaying to the file-backed editor.
Release Notes:
- Fixes “Workspace: Copy Relative Path” not copying while the Find bar
is focused.
This PR adds automatic markdown URL embedding on paste when you are in
text associated with the Markdown language and you have a valid URL in
your clipboard. This the default behavior in VS Code and GitHub, when
pasting a URL in Markdown. It works in both singleton buffers and multi
buffers.
One thing that is a bit unfortunate is that, previously, `do_paste` use
to simply call `Editor::insert()`, in the case of pasting content that
was copied from an external application, and now, we are duplicating
some of `insert()`'s logic in place, in order to have control over
transforming the edits before they are inserted.
Release Notes:
- Added automatic Markdown URL embedding on paste.
---------
Co-authored-by: Cole Miller <53574922+cole-miller@users.noreply.github.com>
Vim mode currently supports `gt` (go to next tab) and `gT` (go to
previous tab) but not with count. Implement the expected behavior as
defined by vim:
- `<count>gt` moves to tab `<count>`
- `<count>gT` moves to previous tab `<count>` times (with wraparound)
Release Notes:
- Improved vim `gt` and `gT` to support count, e.g. `5gt` - go to tab 5,
`8gT` - go to 8th previous tab with wraparound.
This button already exists in the main menu, as well as the "New
Session" view in the debugger panel. However, this view disappears after
starting the debugging session. This PR adds the same button to the
debugger control strip that remains accessible. This is convenient for
people editing their debug.json frequently.
Site-node: I feel like the `Cog` icon would be more appropriate, but I
picked `Code` to stay consistent with the "New Session" view.
Before:
<img width="194" height="118" alt="image"
src="https://github.com/user-attachments/assets/5b42a8a4-f48f-4145-a425-53365dd785ca"
/>
After:
<img width="194" height="118" alt="image"
src="https://github.com/user-attachments/assets/12f56ea1-150b-4564-8e6a-da4671f52079"
/>
Release Notes:
- Added "Edit debug.json" button to debugger control strip
