When using the latest version of `GPUI`, I found some grid layout
issues. I discovered #43555 modified the default behavior of grid
template columns. I checked the implementation at
https://tailwindcss.com/docs/grid-template-columns, and it seems our
previous implementation was correct.
If a grid layout is placed inside a flexbox, the layout becomes
unpredictable.
```rust
impl Render for HelloWorld {
fn render(&mut self, _window: &mut Window, _cx: &mut Context<Self>) -> impl IntoElement {
div()
.flex()
.size(px(500.0))
.bg(rgb(0x505050))
.text_xl()
.text_color(rgb(0xffffff))
.child(
div()
.size_full()
.gap_1()
.grid()
.grid_cols(2)
.border_1()
.border_color(gpui::red())
.children((0..10).map(|ix| {
div()
.w_full()
.border_1()
.border_color(gpui::green())
.child(ix.to_string())
})),
)
}
}
```
| Before | After |
| - | - |
| <img width="612" height="644" alt="After1"
src="https://github.com/user-attachments/assets/64eaf949-0f38-4f0b-aae7-6637f8f40038"
/> | <img width="612" height="644" alt="Before1"
src="https://github.com/user-attachments/assets/561a508d-29ea-4fd2-bd1e-909ad14b9ee3"
/> |
I also placed the grid layout example inside a flexbox too.
| Before | After |
| - | - |
| <img width="612" height="644" alt="After"
src="https://github.com/user-attachments/assets/fa6f4a2d-21d8-413e-8b66-7bd073e05f87"
/> | <img width="612" height="644" alt="Before"
src="https://github.com/user-attachments/assets/9e0783d1-18e9-470d-b913-0dbe4ba88835"
/> |
I tested the changes from the previous PR, and it seems that setting the
table's parent to `v_flex` is sufficient to achieve a non-full table
width without modifying the grid layout. This was already done in the
previous PR.
I reverted the grid changes, the blue border represents the table width.
cc @RemcoSmitsDev
<img width="1107" height="1000" alt="table"
src="https://github.com/user-attachments/assets/4b7ba2a2-a66a-444d-ad42-d80bc9057cce"
/>
So, I believe we should revert to this implementation to align with
tailwindcss behavior and avoid potential future problems, especially
since the cause of this issue is difficult to pinpoint.
Release Notes:
- N/A
Make gpui's `rems()`, `phi()`, `auto()` length related helpers into
const functions.
I can't see why these functions aren't already const except that it
must've been overlooked when they were written?
In my project I had need for rems() to be const, and I thought I'd do
phi() and auto() whilst I was in the neighbourhood
Release Notes:
- N/A
Closes#44148
the existing rate == 0 check inside the timer callback already handles
disabling repeat - it just drops the timer immediately. So the fix
prevents the crash while preserving correct behavior.
Release Notes:
- Linux (Wayland): Fixed a crash that could occur when
`characters_per_second` was zero
Related to https://github.com/zed-industries/zed/pull/44109,
https://github.com/zed-industries/zed/issues/43884,
https://github.com/zed-industries/zed/issues/43809.
In the Linux dispatcher, we create one background thread per CPU, but
when a single core is available, having a single background thread
significantly hinders the perceived performance of Zed. This is
particularly helpful when SSH remoting to low-resource servers.
We may want to bump this to more than two threads actually, but I wanted
to be conservative, and this seems to make a big difference already.
Release Notes:
- N/A
`WorkItemPriority::High` will enqueue the work items to threads with
higher-than-normal priority. If the work items are very intensive, this
can cause the system to become unresponsive. It's not clear what this
gets us, so let's avoid the responsiveness issue by deleting this.
Release Notes:
- N/A
Closes#39152
This PR fixes an issue where we would render Markdown tables full width
based on their container size. We now render tables based on their
content min size, meaning you are still allowed to make the table render
as it was before by making the columns `w_full`.
I had to change the `div()` to `v_flex().items_start()` because this
introduced a weird displaying behavior of the outside table border,
because the grid container was not shrinking due to It was always taking
up the full width of their container.
**Before**
<img width="1273" height="800" alt="Screenshot 2025-11-26 at 14 37 19"
src="https://github.com/user-attachments/assets/2e152021-8679-48c2-b7bd-1c02768c0253"
/>
**After**
<img width="1273" height="797" alt="Screenshot 2025-11-26 at 14 56 12"
src="https://github.com/user-attachments/assets/4459d20e-8c3b-487b-a215-c95ee5c1fc8e"
/>
**Code example**
```markdown
| Name | Age | Occupation |
|:--------:|:-------:|:--------------:|
| Alice | 28 | Engineer |
| Bob | 34 | Designer |
| Carol | 25 | Developer |
| Syntax | Description |
| ----------- | ----------- |
| Header | Title |
| Paragraph | Text |
| City | Population (approx.) | Known For |
|----------------|----------------------|------------------------------------|
| New York | 8,500,000 | Statue of Liberty, Wall Street |
| Los Angeles | 4,000,000 | Hollywood, film industry |
| Chicago | 2,700,000 | Architecture, deep-dish pizza |
| Houston | 2,300,000 | NASA, energy industry |
| Miami | 470,000 | Beaches, Latin culture |
| San Francisco | 800,000 | Golden Gate Bridge, Silicon Valley |
| Las Vegas | 650,000 | Casinos, nightlife |
<table>
<caption>Table Caption</caption>
<thead>
<tr>
<th>ID asjkfjaslkf jalksjflksajflka jlksdla k</th>
<th>Name</th>
</tr>
</thead>
<tbody>
<tr>
<td>1</td>
<td>Chris</td>
</tr>
<tr>
<td>2</td>
<td>Dennis</td>
</tr>
<tr>
<td>3</td>
<td>Sarah</td>
</tr>
<tr>
<td>4</td>
<td>Karen</td>
</tr>
</tbody>
</table>
```
cc @bennetbo
Release Notes:
- Markdown Preview: Markdown tables scale now based on their content
size
Windows applications tend to be fairly re-entrant which does not work
well with our current top-level `RefCell` approach. We have worked
around a bunch of panics in the past due to this, but ultimately this
will re-occur (and still does according to sentry) in the future. So
this PR moves all interior mutability down to the fields.
Fixes ZED-1HM
Fixes ZED-3SH
Fixes ZED-1YV
Fixes ZED-29S
Fixes ZED-29X
Fixes ZED-369
Fixes ZED-20W
Release Notes:
- Fixed panics on windows caused by unexpected re-entrancy for interior
mutability
After #39928, if a font's weight changes between text runs without other
decoration changing, the earlier weight continues to be used for the
subsequent run(s).
PR #40840 fixes this in shape_text, but similar code exists also in
layout_text. The latter is used for text in the editor view itself, so
the issue continues to appear when using a highlighting theme with
varied font weights.
Fix the issue by applying the same fix in layout_text.
Closes#42297
Release Notes:
- Fixed incorrect font weights in editor view when using a highlighting
theme with varying font weights
This reduces hangs on windows when we have many tasks queued up on the
main thread that yield a lot.
Release Notes:
- Reduced hangs on windows in some situations
CLOSES #39042
This is a reopening of #34985+.
_Original descrioption_:
In Wayland, the client implement key repeat themself. In Zed this is
ultimately handled by the gpui crate by inserting a timer source into
the event loop which repeat itself if the key is still held down [1].
But it seems the processing of the repeated key event happen
synchronously inside the timer source handler, meaning the effective
rate become slightly lower (since the repeated timer is scheduled using
the 1/rate as delay).
I measured the event processing time on my laptop and it's typically
around 3ms, but sometimes spiking at 10ms. At low key repeat rates this
is probably not _very_ noticeable. I see the default in Zed is set to a
(measly) 16/s, but I assume most systems will use something closer to
25, which is a 40ms delay. So ~3ms is around 7.5% of the delay. At
higher rate the discrepancy become worse of course.
I can visible notice the spikes, and doing some crude stopwatch
measurements using gedit as a reference I can reproduce around 5-10%
slower rates in Zed.
IMO this is significant enough to warrant improving, especially since
some people can get quite used the repeat rate and might feel something
being "off" in Zed.
~~The suggested fix simply subtract the processing time from the next
delay timer.~~
[1] 32df726f3b/crates/gpui/src/platform/linux/wayland/client.rs (L1355)
Release Notes:
- Improved Wayland (Linux) key repeat rate precision
It is optional for Wayland servers to support server-side decorations.
In particular, GNOME chooses to not implement SSD
(https://gitlab.gnome.org/GNOME/mutter/-/issues/217). So, even if the
application requests SSD, it must draw client-side decorations unless
the application receives a response from the server confirming the
request for SSD.
Before, when the user requested SSD for Zed, but the Wayland server did
not support it, there were no server-side decorations (window titlebar)
drawn, but Zed did not draw the window minimize, maximize, and close
buttons either. This fixes Zed so it always draws the window control
buttons if the Wayland server does not support SSD.
Before on GNOME Wayland with SSD requested:
<img width="3840" height="2160" alt="image"
src="https://github.com/user-attachments/assets/68a6d853-623d-401f-8e7f-21d4dea00543"
/>
After on GNOME Wayland with SSD requested:
<img width="3840" height="2160" alt="image"
src="https://github.com/user-attachments/assets/b258ae8b-fe0e-4ba2-a541-ef6f2c38f788"
/>
Release Notes:
- Fixed window control buttons not showing in GNOME Wayland when SSD
requested
Reverts zed-industries/zed#36848
Turns out this broke copying a screenshot from apps like CleanShot X and
then pasting it over. We should land this again after taking a look at
those cases. Pasting screenshots from the native macOS screenshot
functionality works though.
cc @seantimm
Release Notes:
- Fixed issue where copying a screenshot from apps like CleanShot X into
Agent Panel didn't work as expected.
Problem: Some macOS environments report no devices via
MTLCopyAllDevices, causing startup failure with “unable to access a
compatible graphics device,” especially on Apple Silicon.
Change: Prefer MTLCreateSystemDefaultDevice
(metal::Device::system_default()) first. If None, enumerate devices and
select a non‑removable, low‑power device by preference.
Why this works: On Apple Silicon the system default is the unified GPU;
on Intel, the fallback keeps a stable policy and avoids accidentally
picking removable/high‑power devices.
Impact: Fixes startup on affected ASi systems; improves selection
consistency on Intel multi‑GPU. Behavior unchanged where
system_default() succeeds.
Risk: Low. Aligns with Apple’s recommended selection path. Still fails
early with a clearer message if no Metal devices exist.
Closes#37689.
Release Notes:
- Fixed: Startup failure on some Apple Silicon machines when Metal
device enumeration returned no devices by falling back to the system
default device.
---------
Co-authored-by: 张小白 <364772080@qq.com>
Co-authored-by: Kate <work@localcc.cc>
Part One for Resolving #10910
### Summary
Typing prefix (partial keybinding) will behave like Vim. No timeout
until you either finish the sequence or hit Escape, while ambiguous
sequences still auto-resolve after 1s.
### Description
This follow-up tweaks the which-key system first part groundwork so our
timeout behavior matches Vim’s expectations. Then we can implement the
UI part in the next step (reference latest comments in
https://github.com/zed-industries/zed/pull/34798)
- `DispatchResult` now reports when the current keystrokes are already a
complete binding in the active context stack (`pending_has_binding`). We
only start the 1s flush timer in that case. Pure prefixes or sequences
that only match in other contexts—stay pending indefinitely, so
leader-style combos like `space f g` no longer evaporate after a second.
- `Window::dispatch_key_event` cancels any prior timer before scheduling
a new one and only spawns the background flush task when
`pending_has_binding` is true. If there’s no matching binding, we keep
the pending keystrokes and rely on an explicit Escape or more typing to
resolve them.
Release Notes:
- Fixed multi-stroke keybindings so only ambiguous prefixes auto-trigger
after 1 s; unmatched prefixes now stay pending until canceled, matching
Vim-style leader behavior.
The WM_DPICHANGED suggested RECT is calculated for non-maximized
windows. When a maximized window's DPI changes, we now query the
monitor's work area directly to ensure the window correctly fills the
entire screen.
For non-maximized windows, the original behavior using the
system-suggested RECT is preserved.
Release Notes:
- windows: Fixed maximized window size when DPI scale changes
Signed-off-by: Xiaobo Liu <cppcoffee@gmail.com>
These appear in a lot of stacktraces (especially on windows) despite
them being plain forwarding calls.
Also removes some intermediate calls within gpui that will only turn
into more unnecessary compiler work.
Release Notes:
- N/A *or* Added/Fixed/Improved ...
Calloop (used by our linux executor) was running all futures regardless
of how long they take. Unfortunaly some of our futures are rather busy
and take a while (>10ms).
Running all of them froze the editor for multiple seconds or even
minutes when opening a large project diff (git reset HEAD~2000 in
chromium for example).
Closes #ISSUE
Release Notes:
- N/A
---------
Co-authored-by: Jakub Konka <kubkon@jakubkonka.com>
Closes https://github.com/zed-industries/zed/issues/42944
The powershell we discovered might be in a directory with higher
permission requirements which will cause us to fail using it.
Release Notes:
- Fixed powershell discovery disregarding admin requirements
This PR updates window background rendering to use the **official DWM
backdrop API** (`DwmSetWindowAttribute`) instead of the legacy
`SetWindowCompositionAttribute`.
It also adds **Mica** and **Mica Alt** options to
`WindowBackgroundAppearance` for native Windows 11 effects.
### Motivation
Enables modern, stable, and GPU-accelerated backdrops consistent with
Windows 11’s Fluent Design.
Removes reliance on undocumented APIs while maintaining backward
compatibility with older Windows versions.
### Changes
* Added `MicaBackdrop` and `MicaAltBackdrop` variants.
* Switched to DWM API for applying backdrop effects.
* Verified fallback behavior on Windows 10.
### Release Notes:
- Added `WindowBackgroundAppearance::MicaBackdrop` and
`WindowBackgroundAppearance::MicaAltBackdrop` for Windows 11 Mica and
Mica Alt window backdrops.
### Screenshots
- `WindowBackgroundAppearance::Blurred`
<img width="553" height="354" alt="image"
src="https://github.com/user-attachments/assets/57c9c25d-9412-4141-94b5-00000cc0b1ec"
/>
- `WindowBackgroundAppearance::MicaBackdrop`
<img width="553" height="354" alt="image"
src="https://github.com/user-attachments/assets/019f541c-3335-4c9e-b026-71f5a1786534"
/>
- `WindowBackgroundAppearance::MicaAltBackdrop`
<img width="553" height="354" alt="image"
src="https://github.com/user-attachments/assets/5128d600-c94d-4c89-b81a-8b842fe1337a"
/>
---------
Co-authored-by: John Tur <john-tur@outlook.com>
Closes #ISSUE
Remove unused method `HttpClient::type_name`. Looking at the PR from a
year ago when it was added, it was never actually used for anything and
seems like a prototyping artifact.
Other misc changes for the `http_client` crate include:
- Use `derive_more::Deref` for `HttpClientWithUrl` (already used for
`HttpClientWithProxy`)
- Move `http_client::proxy()` higher up in the trait definition. (It was
in between methods that have default implementations)
Release Notes:
- N/A *or* Added/Fixed/Improved ...
This enables optimizations for our own proc-macros as well as some heavy
hitters. Additionally this gates the `derive_inspector_reflection` to be
skipped for rust-analyzer as it currently slows down rust-analyzer way
too much
Release Notes:
- N/A *or* Added/Fixed/Improved ...
Closes#4947
Taken inspiration from @tasuren implementation, plus the addition for
the double click enabled on the whole title bar too to
maximizes/restores the window.
I was not able to test the application on Linux, no need to test on
Windows since the feature is enabled by the OS.
Release Notes:
- Fixed title bar not fully draggable on macOS
- Fixed not being able to maximizes/restores the window with double
click on the whole title bar on macOS
Summary
Fixes UI scaling issue that occurs when starting Zed after disconnecting
an external monitor on macOS. The window's scale factor and drawable
size are now properly updated when the window changes screens.
Problem Description
When an external monitor is disconnected and Zed is started with only
the built-in screen active, the UI scale becomes incorrect. This happens
because:
1. macOS triggers the `window_did_change_screen` callback when a window
moves between displays (including when displays are disconnected)
2. The existing implementation only restarted the display link but
didn't update the window's scale factor or drawable size
3. This left the window with stale scaling information from the previous
display configuration
Root Cause
The `window_did_change_screen` callback in
`crates/gpui/src/platform/mac/window.rs` was missing the logic to update
the window's scale factor and drawable size when moving between screens.
This logic was only present in the `view_did_change_backing_properties
callback`, which isn't triggered when external monitors are
disconnected.
Solution
- Extracted common logic: Created a new `update_window_scale_factor()`
function that encapsulates the scale factor and drawable size update
logic
- Added scale factor update to screen change: Modified
`window_did_change_screen` to call this function after restarting the
display link
- Refactored existing code: Updated `view_did_change_backing_properties`
to use the new shared function, reducing code duplication
The fix ensures that whenever a window changes screens (due to monitor
disconnect, reconnect, or manual movement), the scale factor, drawable
size, and renderer state are properly synchronized.
Testing
- ✅ Verified that UI scaling remains correct after disconnecting
external monitor
- ✅ Confirmed that reconnecting external monitor works properly
- ✅ Tested that manual window movement between displays updates scaling
correctly
- ✅ No regressions observed in normal window operations
To verity my fix worked I had to copy my preview workspace over my dev
workspace, once I had done this I could reproduce the issue on main
consistently. After switching to the branch with this fix the issue was
resolved.
The fix is similar to what was done on
https://github.com/zed-industries/zed/pull/35686 (Windows)
Closes#37245#38229
Release Notes:
- Fixed: Update scale factor and drawable size when macOS window changes
screen
---------
Co-authored-by: Kate <work@localcc.cc>
This PR introduces `Nearest` scrolling strategy to `UniformList`. This
is now used in completions menu and the picker to choose the appropriate
scrolling strategy depending on movement direction. Previously,
selecting the next element after the last visible item caused the menu
to scroll with `ScrollStrategy::Top`, which scrolled the whole page and
placed the next element at the top. This behavior is inconsistent,
because using `ScrollStrategy::Top` when moving up only scrolls one
element, not the whole page.
https://github.com/user-attachments/assets/ccfb238f-8f76-4a18-a18d-bbcb63340c5a
The solution is to introduce the `Nearest` scrolling strategy which will
internally choose the scrolling strategy depending on whether the new
selected item is below or above currently visible items. This ensures a
single-item scroll regardless of movement direction.
https://github.com/user-attachments/assets/8502efb8-e2c0-4ab1-bd8d-93103841a9c4
I also noticed that some functions in the file have different logic
depending on `y_flipped`. This appears related to reversing the order of
elements in the list when the completion menu appears above the cursor.
This was a feature suggested in #11200 and implemented in #23446. It
looks like this feature was reverted in #27765 and there currently seem
to be no way to have `y_flipped` to be set to `true`.
My understanding is that the opposite scroll strategy should be used if
`y_flipped`, but since there is no way to enable this feature to test it
and I don't know if the feature is ever going to be reintroduced I
decided not to include it in this PR.
Release Notes:
- gpui: Add 'Nearest' scrolling strategy to 'UniformList'
