`Subscription::Notification` held a strong `Arc` to a language server's
entire notification handler map, while `Subscription::Io` already used a
`Weak`. Any subscription retained beyond the server's lifetime (e.g.
`copilot_log_subscription` in the LSP log store, which lives until the
project is closed) kept **every** handler on the dead server alive —
along with everything those closures capture, including
`Arc<CachedLspAdapter>` and `LspAdapterDelegate`s that pin full worktree
snapshots. Each server restart while such a subscription existed leaked
one server generation; on large worktrees that's tens of MB per restart.
The fix changes `Subscription::Notification` to hold a `Weak` reference,
matching the `Io` variant. Dropping a subscription still deregisters its
handler when the server is alive; if the server is already gone, there
is nothing left to clean up.
Also adds a regression test
(`test_subscription_leaks_handlers_after_server_drop`) to prevent this
from happening again
Helps with #58190 and #31461 (long-running memory growth involving
language servers), though it likely isn't the whole story for either,
the unbounded outbound message channel in `crates/lsp/src/lsp.rs`
remains a separate suspect for the extreme growth in #58190.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content adheres to Zed's UI standards
([UX/UI](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
and
[icon](https://github.com/zed-industries/zed/blob/main/crates/icons/README.md)
guidelines)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Fixed a memory leak where restarting a language server could leak
memory if its notification subscriptions were still retained (e.g. while
the LSP log view was tracking it).
The background task reading a language server's stdout pushed every
parsed message into an unbounded channel consumed on the foreground
thread. If the main thread stalled while a server kept emitting
notifications, messages accumulated at wire speed with no limit —
matching the report in #58190 of 391 GB of memory growth during a
multi-hour hang with a chatty server.
This bounds the queue at 128 messages. When full, the reader stops
reading the server's stdout, so the OS pipe applies back pressure to the
server instead of Zed buffering its output in memory. No messages are
dropped, and healthy operation is unaffected. In flight client requests
behind a full queue fail via their existing background executor
timeouts.
Adds a regression test that wedges the consumer, asserts the queue stays
bounded, then drains it and asserts no messages were lost.
Helps with #58190 and may also help with #31461.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content adheres to Zed's UI standards
([UX/UI](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
and
[icon](https://github.com/zed-industries/zed/blob/main/crates/icons/README.md)
guidelines)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- lsp: Improve Zed's memory usage when LSP's emitted messages faster
than the foreground thread can handle them
This is an unsolicited contribution, I hope that's ok. The feature is
small and I didn't want to open a discussion just for this.
There was no way to write to `.git/info/exclude` from Zed, so I added
one.
The "Add to .gitignore" action in the project panel and git panel
context menus is now grouped under a "Git" submenu, with a new "Add to
.git/info/exclude" action next to it.
Both actions:
- skip the write if the pattern is already there
- make sure there's a trailing newline
- show an error toast if something goes wrong
- fail gracefully on remote repositories
In the git panel both entries are disabled unless the file is untracked,
same as the gitignore action was before.
The write logic is shared via a small private helper, covered by two
unit tests.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Added `Add to .git/info/exclude` option under a new `Git` submenu in
the project panel and git panel context menus, alongside the existing
`Add to .gitignore` action.
---------
Co-authored-by: Cole Miller <cole@zed.dev>
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
The non-recursive Linux watcher causes subdirectories inside `.git/` to
need explicit watching. For repos using the reftable backend, all ref
data lives in `.git/reftable/`. Operations in the UI like uncommitting
and creating new branches only modify refs inside `.git/reftable`
without changing `.git/HEAD`'s contents, so no inotify event fires on
`.git/` itself and the git UI never refreshes.
Essentially all this change does is check if `.git/reftable/` exists and
if it does, add it to the watcher. On the recursive macOS/Windows
watchers, `FsWatcher::add` returns early safely so this *should* not be
a performance loss apart from the brief lookup.
Release Notes:
- Fixed Git UI not refreshing on Linux for repositories using the
reftable backend.
This PR tackles how we communicate errors/warnings relative to skills.
It touches three major surfaces: callouts in the thread view, menu items
in the completion menu, and settings UI. These are all the places where
we display something to tell the user a skill might have a problem.
In the settings UI, we now clamp the skill description to five lines.
And if there's a warning, we show the warning icon and the description
in a tooltip when hovering it:
<img width="600" alt="Screenshot 2026-06-08 at 10 51@2x"
src="https://github.com/user-attachments/assets/1a1233b0-c24a-4082-9fe6-2ab1d00ba9fd"
/>
In the thread view, we display callouts with the warnings. I simplified
the long-description one specifically given we can click on each skill
item right there, so no need for a "manage global skills" button or
descriptions in each item:
<img width="600" alt="Screenshot 2026-06-08 at 10 52@2x"
src="https://github.com/user-attachments/assets/4b9aad23-7362-40cd-8ab5-6dfceb46badd"
/>
In the code completion menu, I removed the icon that appeared at the far
right and instead added the ability for it to render a colored warning
icon instead of the icon it'd already display in the left; I think this
simplifies it nicely:
<img width="350" alt="Screenshot 2026-06-08 at 10 53@2x"
src="https://github.com/user-attachments/assets/8d0c731a-31a0-4d30-abab-1eb40f61031e"
/>
Release Notes:
- Agent: Improved how we communicate problems relative to skills, in
both the agent panel and the settings UI.
---------
Co-authored-by: Martin Ye <martin@zed.dev>
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior (visual difference only)
- [x] Performance impact has been considered and is acceptable
I was unable to find a specific issue or PR that mentions this.
As Linux installations always use built-in prompts and ignore the `Use
System Prompts` setting (as far as I'm aware) all users using Zed (with
client-side decorations) are affected. The darkened prompt backdrop was
drawn across the full window boundary, including the transparent
decoration/shadow inset, causing it to overlap.
This PR fixes the backdrop to respect the client-side decoration inset
and only applies rounded corners on non-tiled edges. Tiled edges remain
flush, matching the rest of the client-side window decoration behavior.
I've verified it works on Wayland, X11 and causes no regressions on
macOS or Windows (with `Use System Prompts` off).
Notice the screenshots below on a white background showing the before
and after, where the faint overlap is shown broken and fixed (sorry if
you have to squint).
Before:
After:
Release Notes:
- Fixed prompt darkened backdrop/background rendering outside
client-decorated window boundaries on Linux.
---------
Co-authored-by: Smit Barmase <heysmitbarmase@gmail.com>
Closes AI-344
Closes AI-342
Moves agent skills management out of the agent panel and into the
settings UI. The skill creator (previously its own crate) now lives in
`settings_ui` as a settings page, alongside the skills setup page. The
agent panel's ellipsis menu now links to the settings page instead of
hosting skill management directly.
Release Notes:
- Improved agent skills management by moving it into the settings UI
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes#58543 and extends to all instances of "Ok" button labels.
Release Notes:
- N/A
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Nucleo tracking issue:
https://github.com/zed-industries/zed/issues/55872
This PR switches the outline picker to use `fuzzy_nucleo`.
with that I refactored the outline fuzzy picking to take advantage of
nucleo and its multi-atom query support. The previous implementation had
a lot going on specifically to work around the lack of multiple atoms,
basically taking matters into its own hands to accomplish the same goals
within the constraints it had.
Instead of having two lists of candidates that we have to run the query
against, we just run a single query and take advantage of the fact that
nucleo chooses the matches that are the furtherest towards the end of
the haystack to implement the same ancestor/leaf filtering. It retains
the leaf only matching on single atom queries.
video:
https://github.com/user-attachments/assets/64baa8d7-fd77-452c-86d1-e08561422d85
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- outline: switch to fuzzy_nucleo
This PR updates the database schema for Collab tests.
The corresponding database schema migration has been created in the
Cloud repo and applied to the production database.
Release Notes:
- N/A
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes #ISSUE
Release Notes:
- N/A or Added/Fixed/Improved ...
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes #ISSUE
Release Notes:
- N/A or Added/Fixed/Improved ...
Closes FR-44
Unlike X11, where Zed already times out clipboard reads, on Wayland we
had no timeout at all. We read the clipboard from a pipe the source app
writes into, and we were doing a blocking `read_to_end` on it. So if the
other app opened the pipe but never wrote anything, or stalled partway
through, the read would block forever and hang Zed.
This PR replaces it with a non-blocking read driven by `poll` with a 4s
timeout, so:
- a stalled writer fails cleanly instead of freezing us.
- a slow writer that's still making progress on a large payload keeps
working.
This roughly mirrors what the X11 path already does. I have tested
pasting text, large image, drag-n-drop, etc cases.
To Reproduce:
1. Copy text from a web page in Firefox.
2. `kill -STOP "$(pgrep -o firefox)"`
3. Paste into Zed.
Zed hangs indefinitely.
<img width="400"
src="https://github.com/user-attachments/assets/532d1c55-1500-4143-8227-127e0024efba"
/>
Release Notes:
- Fixed a freeze on Linux Wayland when reading the clipboard from a slow
or unresponsive application.
This got produced by verifiying ssh askpass things a last time before
merging which with a stale cli exe piped into the prompt as a filename
...
Release Notes:
- N/A or Added/Fixed/Improved ...
Neither really panned out, removing for now.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- N/A
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes#58699
Release Notes:
- N/A
On Windows, Zed generated a .ps1 askpass script and set **SSH_ASKPASS**
to 'powershell.exe -ExecutionPolicy Bypass -File ...'. SSH calls exec()
on SSH_ASKPASS which cannot exec a command string, causing:
error: ssh_askpass: exec(powershell.exe ...): No such file or directory
Fix by pointing **SSH_ASKPASS** directly to cli.exe and passing the
socket path via **ZED_ASKPASS_SOCKET** env var, which cli.exe reads
before clap parses arguments.
---
> Compiling Zed almost blew up my computer.
---
Related #29048
Release Notes:
- Fixed some ssh issues on windows that prevented from connecting to a
remote
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes#55279
When auto-updating, Zed first downloads the update and then performs the
installation. On each platform, this installation phase runs external
commands that spawn new processes. These operations can block the thread
they run on—especially under heavy system load or when
security/antivirus software is active. (An analysis of the blockages on
Windows is documented in
https://github.com/zed-industries/zed/issues/55279#issuecomment-4638072228).
This PR moves the installation process to a background thread. This
ensures that even if process spawning block, the main UI thread remains
responsive.
Release Notes:
- Fixed a UI freeze that could occur when installing auto-updates.
---------
Co-authored-by: Lukas Wirth <me@lukaswirth.dev>
Closes#56778
## Problem
The agent-profile tool configuration lists every built-in tool from
`agent::ALL_TOOL_NAMES`, filtering only by provider support
(`crates/agent_ui/src/agent_configuration/manage_profiles_modal.rs`).
But `Thread::enabled_tools` additionally drops feature-flag-gated tools
before they reach the model — the LSP tools (`go_to_definition`,
`find_references`, `get_code_actions`, `apply_code_action`) behind
`LspToolFeatureFlag`, `rename_symbol` behind `RenameToolFeatureFlag`,
and `create_thread` / `list_agents_and_models` / `update_plan` /
`update_title` behind their own flags.
The result: when a flag is off, a user can enable the tool in a profile,
but the agent never receives it. The picker and the runtime gate
disagreed.
## Fix
Extract the feature-flag gating into a single shared predicate,
`agent::tool_feature_flag_enabled`, and route **both**
`Thread::enabled_tools` and the profile tool picker through it. The
configuration UI can no longer offer a tool the agent would silently
drop, and the two call sites can't drift apart. Also updated the "silent
drop gates" comment in `tools.rs` to document this third gate.
This matches the behavior requested in the issue (don't list tools that
can't be used). The alternative — showing them disabled with an
explanation — is also possible; happy to switch if preferred.
## Testing
- Added `lsp_tools_are_gated_by_their_feature_flag` in
`crates/agent/src/tools.rs`.
- `cargo test -p agent`, `cargo clippy -p agent`, and `cargo fmt` are
clean.
- (`agent_ui` doesn't build in my local environment due to an unrelated
`webrtc-sys` C++ toolchain issue; the `agent_ui` change is a one-line
filter addition mirroring the adjacent provider-support check — relying
on CI to confirm.)
Release Notes:
- Fixed feature-flag-gated tools (such as the LSP tools) being listed in
agent profile configuration when they could not actually be used by the
agent.
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Fixes hardcoded SHA lengths not using the `SHORT_SHA_LENGTH` constant.
Release Notes:
- Reduced short commit SHAs in blame and commit tooltips from 8 to 7 -
consistent with the graph and history views.
This PR bumps the extension CLI version used in the extension workflows
to `9ee3c503a4bbbc6b4a0f8a789acca4871d773223`.
Release Notes:
- N/A
Co-authored-by: zed-zippy[bot] <234243425+zed-zippy[bot]@users.noreply.github.com>
This migrates the `Workspace::show_error` method to take a generic type
that implements `WorkspaceError` instead. That trait containts methods
to show a proper error message but most importantly means to help with
providing actions given certain errors.
The intention for this is rather easy: Whenever we show an error to the
user on the workspace level, that error should be
- actionable to a degree, even if that is only opening logs or whatnot
- show a message that is understandable to the user
- for not-too-critical errors, should hide after a certain delay and not
force the user to dismiss it using their mouse.
For now, there exist trait implementations for string error types and
anyhow errors. The long term goal here is to get rid of these altogether
and migrate everything to be strongly typed and especially actionable
instead.
Self-Review Checklist:
- [ ] I've reviewed my own diff for quality, security, and reliability
- [ ] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [X] Performance impact has been considered and is acceptable
Release Notes:
- N/A
---------
Co-authored-by: Kirill Bulatov <kirill@zed.dev>
Co-authored-by: Ben Brandt <benjamin.j.brandt@gmail.com>
This will improve local dev installation times as well as CI times by a
lot (I measured it once and for some extensions, it might save us up to
60% of CI time. It will similarly help locally).
I also moved the debug adapter schema validation to the other
validations, as it felt it makes more sense to check there as opposed to
doing this during the compilation step.
Release Notes:
- Improved dev extension installation times for language extensions with
multiple grammars or language servers.
Parse git blame stdout incrementally while the process is still running
instead of waiting for child output to buffer the entire command result.
Release Notes:
- N/A or Added/Fixed/Improved ...
Self-Review Checklist:
- [X] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [ ] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [ ] Performance impact has been considered and is acceptable
Fix an issue introduced by PR #58237 where IME could not be enabled when
multiple windows were open.
The fix ensures that `update_ime_enabled` only updates IME state for the
active window.
Release Notes:
- N/A
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [ ] Unsafe blocks (if any) have justifying comments
- [ ] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [ ] Performance impact has been considered and is acceptable
Please check
https://github.com/zed-industries/zed/issues/56277#issuecomment-4417185748
for more context.
Release Notes:
- N/A or Added/Fixed/Improved ...
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Closes #ISSUE
Release Notes:
- N/A or Added/Fixed/Improved ...
We had a nested spawn that was disguising the fact that the actual
`git_config` task was being dropped immediately instead of detached.
This only seems to cause problems in release builds due to timing
issues.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Fixed a bug where clicking the button to trust a git repository would
do nothing.
This PR bumps the version of the GLSL extension to v0.2.4.
Release Notes:
- N/A
Co-authored-by: zed-zippy[bot] <234243425+zed-zippy[bot]@users.noreply.github.com>
Add C-Style indentation rules for GLSL. By adding an indents.scm file
for GLSL we get the same automatic indentation as we have for other
languages.
Release Notes:
- N/A
In the current Python Tree-sitter `highlights.scm`, exception
(https://docs.python.org/3/library/exceptions.html) are highlighted
under `@type.class` which is inconsistent compared to other editors (VS
Code, Neovim, Helix).
This PR adds `#any-of?` with a list of python's exceptions and highlight
them under `@type.builtin`.
| Before | After |
|
:---------------------------------------------------------------------------------------------------------------------------------:
|
:--------------------------------------------------------------------------------------------------------------------------------:
|
| <img width="1237" height="1160" alt="image"
src="https://github.com/user-attachments/assets/05e1ae34-7637-413a-9785-ad696c2bf751"
/> | <img width="1237" height="1154" alt="image"
src="https://github.com/user-attachments/assets/105c61ca-b6a7-4c26-acc8-3e300f044510"
/> |
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Added the ability to stylize builtin python exceptions and warnings
using the `type.class.builtin` syntax capture.
Closes AI-358
This PR hides the chevron and removes hover styles from the project
header when there is an active search, given groups can't be collapsed
while that's happening. We could allow it, but I think it's not super
worth it; this feels like like a pragmatic move.
Release Notes:
- Sidebar: Fixed a bug where affordances to collapse the project header
would appear when searching.
The settings UI previously required selecting a settings scope (User /
project) before entering a sub-page; the scope picker disappeared once
inside. This makes the scope switchable from within sub-pages.
- The scope segment of the sub-page breadcrumb (e.g. "User / AI / Agent
Configuration / Skills") is now a dropdown listing all scopes the
sub-page supports (filtered by the sub-page's file mask, intersected
across the sub-page stack). Selecting a scope re-renders the sub-page in
place — e.g. the Skills page swaps between global and project-local
skills.
- The sub-page stack is preserved across scope changes when every
sub-page in it is valid in the new scope; otherwise it falls back to the
regular navigation behavior.
- Sub-pages that only support one scope (Tool Permissions, Configure
Providers, Feature Flags) keep the static label.
- Entering a sub-page now clears the settings search bar so sub-pages
show all their items (settings deep links that prefill the search are
unaffected).
Closes AI-343
Release Notes:
- Improved the settings window: sub-pages such as Skills and
per-language settings now have a scope picker in the breadcrumb, so you
can switch between User and project settings without leaving the page.
---------
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
For the longest time, we've had a labeled "Share" button on the title
bar whenever you're in a call. However, it was extremely easy to click
that button thinking you'd be sharing _your screen_ instead of your
project. At the end of the day, there wasn't any specific cue that'd
allow you to know what you'd be sharing. This is why people just assumed
you'd share the screen, which is arguably way more common than sharing
the project. We've seen users bump into this confusion countless times,
and this also even happens to us (the whole Zed staff) weekly whenever
we're in meetings.
So, this PR changes that button to be an icon button instead, using a
folder icon to better communicate it means sharing the project. I also
took advantage of the opportunity to change the approach we use for
adding these buttons, using `when` instead of `children.push`; I think
the former is just easier to understand, allowing to more easily picture
the UI in your (my) head. Lastly, also improved what we display in the
tooltip for the project share button, by showing the names of all
projects you'll be sharing, as well as the tooltip that displayed call
diagnostics.
| Before | After |
|--------|--------|
| <img width="778" height="360" alt="Screenshot 2026-06-05 at 3 48
2@2x"
src="https://github.com/user-attachments/assets/363593fb-79f8-42e9-83b9-345bc731846f"
/> | <img width="778" height="360" alt="Screenshot 2026-06-05 at 3 48
3@2x"
src="https://github.com/user-attachments/assets/32afc412-a1fb-45a4-a19e-74e4923c0c8b"
/> |
Release Notes:
- Improved call controls design, making it clearer what button shares
_the project_ vs. _the screen_.
## Context
`apply_remote_edit` and `apply_local_edit` rebuild the fragment
`SumTree` through `FragmentBuilder`. Since #51941, the builder collapsed
the entire tree into a flat `Vec<Fragment>` (cloning every fragment) and
rebuilt a fresh tree from scratch on every call. The new tree shared no
nodes with the previous one, so each edit cost O(N) in the *total*
number of fragments: clone all fragments, allocate an entirely new tree,
and then free the whole previous tree on assignment — the last part
showing up as a large amount of time spent dropping `SumTree`s.
That batching was a deliberate win for the bulk `replace_all` path
(#51941, one `edit()` carrying millions of ranges), but it penalizes
every other caller — most notably `apply_remote_edit`, which runs once
per op in a loop in `apply_ops` and so rebuilt the whole tree per remote
operation.
## This change
Make `FragmentBuilder` chunk-based. Appended slices are kept as intact
`SumTree` subtrees, so they keep sharing nodes with the previous tree;
only individually pushed fragments are batched into `Vec`s.
`to_sum_tree` appends the shared subtrees (touching just the right
spine) and builds the loose runs in one pass, parallelizing the large
ones.
Net effect:
- Small edits on large/heavily-fragmented buffers (the
`apply_remote_edit` collaborative path) go back to O(edited + log N)
with a cheap drop, instead of O(total fragments).
- The bulk `replace_all` path keeps its batched build and is not
regressed.
## Benchmarks
Using the file from #38927 (619 MB CSV, 10,325,246 matches of `"` →
`""`), `release-fast`:
| Case | flatten (#51941) | this PR |
|---|---|---|
| `replace_all` (all matches) | 50.70 s | 45.16 s |
| replace 1 match (`--single`) | 16.0 ms | 16.0 ms |
`replace_all` is ~11% faster and not regressed. The
single-match-on-a-freshly-loaded-file case is unchanged, because a fresh
`Buffer::local` is barely fragmented, so even the from-scratch rebuild
is cheap there; the structural-sharing win is on heavily-fragmented
buffers receiving small edits.
The `--single` flag added to `editor_benchmarks` makes the latter case
measurable.
Release Notes:
- Improved the performance of applying edits to large buffers
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
In a multi-workspace window several workspaces share one platform
window. Switching the active workspace did not update the window title
or edited indicator, and background workspaces could clobber the active
workspace's title via their still-live project/item event subscriptions.
Track the active workspace id in a shared Rc<Cell<EntityId>> owned by
the MultiWorkspace and read by each member workspace, so a workspace
only writes the shared window's title/edited state when it is the active
one. The newly active workspace re-applies its title and edited
indicator on activation. We use a Rc-cell here to avoid double lease
issues as we cannot tell whether we are inside a running MultiWorkspace
lease or not in the given codepaths.
Release Notes:
- Fixed the window title not updating when switching between workspaces
in a multi-workspace window
---------
Co-authored-by: Eric Holk <eric@theincredibleholk.org>
When a pending tool call awaiting permission contains tall content (e.g.
a full plan in Claude Code plan mode) and the inline prompt is scrolled
out of view, the floating awaiting-permission row embeds that content
and can grow to consume the entire panel, squeezing the conversation
list to zero height. `ListState::item_is_above_viewport` /
`item_is_below_viewport` returned `None` for a zero-height viewport, so
the row's visibility decision oscillated from frame to frame, flickering
between the conversation and the permission prompt and making the thread
unusable.
This PR fixes the root cause and hardens the UI:
- `gpui`: `item_is_above_viewport` / `item_is_below_viewport` now answer
definitively from the last layout bounds even when the viewport is
zero-height, so callers that size sibling UI based on these queries
can't oscillate. Adds a regression test.
- `agent_ui`: adds a `Floating` tool call layout used by the
awaiting-permission row that caps the embedded tool call content with a
scrollable max height, so the row can never crowd the conversation list
out of view while keeping the permission buttons visible. Adds an
integration test that drives real window draws and asserts the row's
visibility is stable across frames.
Closes AI-372
Closes https://github.com/zed-industries/zed/issues/58564
Release Notes:
- Fixed the agent panel flickering and becoming unusable when a
permission prompt with tall content (such as a plan awaiting approval)
was scrolled out of view.
This crash occurred when the left hand side (LHS) editor in the split
diff retained a stale shared scroll handler after the split editor
unsplits. I added an invariant check to our randomized companion display
map tests to prevent regressions here and explicitly make LHS editor set
a native scroll anchor when unsplitting.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Fix crash that could occur when unsplitting a side by side diff view
Stabilizes context windows + persistent session/delete for acp agents
that support them
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- acp: Show context window usage and cost metrics for external agents
that support it.
- acp: Persist session deletion from history for external agents that
support it.
Closes#55170
Fixes Helix mode undo selection history after `d` deletes the character
under an empty cursor. Previously, `d` temporarily expanded the cursor
to a one-character selection before deleting, and undo restored that
temporary selection. When the deleted character was a newline, pressing
`x` after undo selected additional line which was wrong.
This preserves the original Helix selection as the undo-restored
selection while keeping the existing delete behavior unchanged.
Self-Review Checklist:
- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [x] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable
Release Notes:
- Fixed Helix mode selecting one extra line after undoing a newline
delete
Co-authored-by: Tom Houlé <13155277+tomhoule@users.noreply.github.com>
Anthony Eid