Currently when we render text with differing styles adjacently we might
form a ligature between the text, causing the ligature forming
characters to take on one of the two styles. This can especially become
confusing when a ligature is formed between actual text and inlay hints.
Annoyingly, the only ways to prevent this with core text is to either
render each run separately, or to insert a zero-width non-joiner to
force core text to break the ligatures apart, as it otherwise will merge
subsequent font runs of the same fonts.
We currently do layouting on a per line basis and it is unlikely we want
to change that as it would incur a lot of complexity and annoyances to
merge things back into a line, so this goes with the other approach of
inserting ZWNJ characters instead.
Note that neither linux nor windows seem to currently render ligatures,
so this only concerns macOS rendering at the moment.
Release Notes:
- Fixed ligatures forming between real text and inlay hints on macOS
Follow up of https://github.com/zed-industries/zed/pull/33334
After chatting with @MrSubidubi we found out that he had an old defaults
setting (most likely from when he encountered a previous window tabbing
bug):
```
❯ defaults read dev.zed.Zed-Nightly
{
NSNavPanelExpandedSizeForOpenMode = "{800, 448}";
NSNavPanelExpandedSizeForSaveMode = "{800, 448}";
NSNavPanelExpandedStateForSaveMode = 1;
NSOSPLastRootDirectory = {length = 828, bytes = 0x626f6f6b 3c030000 00000410 30000000 ... dc010000 00000000 };
"NSWindow Frame NSNavPanelAutosaveName" = "557 1726 800 448 -323 982 2560 1440 ";
"NSWindowTabbingShoudShowTabBarKey-GPUIWindow-GPUIWindow-(null)-HT-FS" = 1;
}
```
> That suffix is AppKit’s fallback autosave name when no tabbing
identifier is set. It encodes the NSWindow subclass (GPUIWindow), plus
traits like HT (hidden titlebar) and FS (fullscreen).
Which explains why it only happened on the Nightly build, since each
bundle has it's own defaults. It also explains why the tabbar would
disappear when he activated the `use_system_window_tabs` setting,
because with that setting activated, the tabbing identifier becomes
"zed" (instead of the default one when omitted) for which he didn't have
the `NSWindowTabbingShoudShowTabBarKey` default.
The original implementation was perhaps a bit naive and relied fully on
macOS to determine if the tabbar should be shown. I've updated the code
to always hide the tabbar, if the setting is turned off and there is
only 1 tab entry.
While testing, I also noticed that the menu's like 'merge all windows'
wouldn't become active when the setting was turned on, only after a full
workspace reload. So I added a setting observer as well, to immediately
set the correct window properties to enable all the features without a
reload.
Release Notes:
- N/A
This pull request refactors the `KeybindingKeystroke` struct and related
code to improve platform abstraction. The changes centralize
platform-specific logic within `KeybindingKeystroke` and update its
usage throughout the codebase, making the API more consistent and less
error-prone.
Release Notes:
- N/A
Closes#36300
This PR follows Windows conventions by introducing
`KeybindingKeystroke`, so shortcuts now show up as `ctrl-shift-4`
instead of `ctrl-$`.
It also fixes issues with keyboard layouts: when `use_key_equivalents`
is set to true, keys are remapped based on their virtual key codes. For
example, `ctrl-\` on a standard English layout will be mapped to
`ctrl-ё` on a Russian layout.
Release Notes:
- N/A
---------
Co-authored-by: Kate <kate@zed.dev>
This removes around 900 unnecessary clones, ranging from cloning a few
ints all the way to large data structures and images.
A lot of these were fixed using `cargo clippy --fix --workspace
--all-targets`, however it often breaks other lints and needs to be run
again. This was then followed up with some manual fixing.
I understand this is a large diff, but all the changes are pretty
trivial. Rust is doing some heavy lifting here for us. Once I get it up
to speed with main, I'd appreciate this getting merged rather sooner
than later.
Release Notes:
- N/A
Closes#36229
Fix zero-sized texture creation that triggers a SIGABRT in the Metal
renderer. Not sure why this happens yet, but it likely occurs when
`native_window.contentView()` returns a zero `NSSize` during initial
window creation, before the view size is computed.
Release Notes:
- Fixed a rare startup crash on macOS.
Closes#24672
This PR complements a feature added earlier by @JosephTLyons (in
https://github.com/zed-industries/zed/pull/32353) where the text is
considered as the tab title in a new buffer. It piggybacks off that
change and sets the title as the suggested filename in the save dialog
(completely mirroring the same functionality in VSCode):
Release Notes:
- Text entered in a new untitled buffer is considered as the default
filename when saving
The Zed Plex fonts were found to violate the OFL by using the word Plex
in the name.
Lilex has better ligatures and box-drawing characters than Zed Plex
Mono, but Zed Plex Sans should be identical
to IBM Plex Sans.
Closes#15542Closeszed-industries/zed-fonts#31
Release Notes:
- The "Zed Plex Sans" and "Zed Plex Mono" fonts have been replaced with
"IBM Plex Sans" and "Lilex". The old names still work for backward
compatibility. Other than fixing line-drawing characters, and improving
the ligatures, there should be little visual change as the fonts are all
of the same family.
- Introduced ".ZedSans" and ".ZedMono" as aliases to allow us to easily
change the default fonts in the future. These currently default to "IBM
Plex Sans" and "Lilex" respectively.
Release Notes:
- N/A
---
In the same vein as #29538, the "Services" menu on macOS depended on the
text being exactly "Services", not allowing for i18n of the menu name.
This PR introduces a new menu type called `OsMenu` that defines a
special menu that can be populated by the system. Currently, it takes
one enum value, `ServicesMenu` that tells the system to populate its
contents with the items it would usually populate the "Services" menu
with.
An example of this being used has been implemented in the `set_menus`
example:
`cargo run -p gpui --example set_menus`
---
Point to consider:
In `mac/platform.rs:414` the existing code for setting the "Services"
menu remains for backwards compatibility. Should this remain now that
this new method exists to set the menu, or should it be removed?
---------
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
Fixes a regression introduced in
https://github.com/zed-industries/zed/pull/34992
### Background
Paths are rendered first to an intermediate MSAA texture, and then
copied to the final drawable. Because paths can have transparency, it's
important that pixels are not copied repeatedly if paths have
overlapping bounding boxes. When N paths have the same draw order, we
infer that they must have disjoint bounding boxes, so that we can copy
them each individually (as opposed to copying a single rect that
contains them all). Previously, the bounding box that we were using to
copy paths was not accounting for the path's content mask (but it is
accounted for in the bounds tree that determines their draw order).
This cause bugs like this, where certain path pixels spuriously had
their opacity doubled:
https://github.com/user-attachments/assets/d792e60c-790b-49ad-b435-6695daba430f
This PR fixes that bug.
* [x] mac
* [x] linux
* [x] windows
Release Notes:
- Fixed a bug where a selection's opacity was computed incorrectly when
it overlapped with another editor's selections in a certain way.
I was interested in potentially using gpui for a hobby project, but
needed [layer
shell](https://wayland.app/protocols/wlr-layer-shell-unstable-v1)
support for it. Turns out gpui's (excellent!) architecture made that
super easy to implement, so I went ahead and did it.
Layer shell is a window role used for notification windows, lock
screens, docks, backgrounds, etc. Supporting it in gpui opens the door
to implementing applications like that using the framework.
If this turns out interesting enough to merge - I'm also happy to
provide a follow-up PR (when I have the time to) to implement some of
the desirable window options for layer shell surfaces, such as:
- namespace (currently always `""`)
- keyboard interactivity (currently always `OnDemand`, which mimics
normal keyboard interactivity)
- anchor, exclusive zone, margins
- popups
Release Notes:
- Added support for wayland layer shell surfaces in gpui
---------
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
This is another attempt to solve the same problem as
https://github.com/zed-industries/zed/pull/29718, while avoiding the
regression on Intel GPUs.
### Background
Currently, on main, all paths are first rendered to an intermediate
"atlas" texture, similar to what we use for rendering glyphs, but with
multi-sample antialiasing enabled. They are then drawn into our actual
frame buffer in a separate pass, via the "path sprite" shaders.
Notably, the intermediate texture acts as an "atlas" - the paths are
laid out in a non-overlapping way, so that each path could be copied to
an arbitrary position in the final scene. This non-overlapping approach
makes a lot sense for Glyphs (which are frequently re-used in multiple
places within a frame, and even across frames), but paths do not have
these properties.
* we clear the atlas every frame
* we rasterize each path separately. there is no deduping.
The problem with our current approach is that the path atlas textures
can end up using lots of VRAM if the scene contains many paths. This is
more of a problem in other apps that use GPUI than it is in Zed, but I
do think it's an issue for Zed as well. On Windows, I have hit some
crashes related to GPU memory.
In https://github.com/zed-industries/zed/pull/29718, @sunli829
simplified path rendering to just draw directly to the frame buffer, and
enabled msaa for the whole frame buffer. But apparently this doesn't
work well on Intel GPUs because MSAA is slow on those GPUs. So we
reverted that PR.
### Solution
With this PR, we rasterize paths to an intermediate texture with MSAA.
But rather than treating this intermediate texture like an *atlas*
(growing it in order to allocate non-overlapping rectangles for every
path), we simply use a single fixed-size, color texture that is the same
size as thew viewport. In this texture, we rasterize the paths in their
final screen position, allowing them to overlap. Then we simply blit
them from the resolved texture to the frame buffer.
### To do
* [x] Implement for Metal
* [x] Implement for Blade
* [x] Fix content masking for paths
* [x] Fix rendering of partially transparent paths
* [x] Verify that this performs well on Intel GPUs (help @notpeter 🙏 )
* [ ] Profile and optimize
Release Notes:
- N/A
---------
Co-authored-by: Junkui Zhang <364772080@qq.com>
Instead of selecting a screen to share arbitrarily, we'll now allow user
to select the screen to share. Note that sharing multiple screens at the
time is still not supported (though prolly not too far-fetched).
Related to #4666
Release Notes:
- Added screen selector dropdown to screen share button
---------
Co-authored-by: Kirill Bulatov <kirill@zed.dev>
Co-authored-by: Cole Miller <cole@zed.dev>
Reverts zed-industries/zed#29718
We've noticed some issues with Zed on Intel-based Macs where typing has
become sluggish, and git bisect has seemed to point towards this PR.
Reverting for now, until we can understand why it is causing this issue.
Closes#22869
Release Notes:
- Added `show_menus` setting to always show menu bar for Linux and
Windows.
---------
Co-authored-by: Cole Miller <cole@zed.dev>
Co-authored-by: Danilo Leal <daniloleal09@gmail.com>
Currently, the rendering path required creating a texture for each path,
which wasted a large amount of video memory. In our application, simply
drawing some charts resulted in video memory usage as high as 5G.
I removed the step of creating path textures and directly drew the paths
on the rendering target, adding post-processing global multi-sampling
anti-aliasing. Drawing paths no longer requires allocating any
additional video memory and also improves the performance of path
rendering.
Release Notes:
- N/A
---------
Co-authored-by: Jason Lee <huacnlee@gmail.com>
Add `screen-capture` feature to gpui to enable screen capture support. The motivation for this is to make dependencies on scap / x11 / xcb optional.
Release Notes:
- N/A
---------
Co-authored-by: Michael Sloan <michael@zed.dev>
Closes#15383Closes#10993
`NSVisualEffectView` is an official API for implementing blur effects
and, by traversing the layers, we **can remove the background color**
that comes with the view. This avoids using private APIs and aligns
better with macOS’s native design.
Currently, `GPUIView` serves as the content view of the window. To add
the blurred view, `GPUIView` is downgraded to a subview of the content
view, placed at the same level as the blurred view.
Release Notes:
- Fixed the missing shadow for blurred-background windows on macOS.
---------
Co-authored-by: Peter Tripp <peter@zed.dev>
Closes#21700
Release Notes:
- Added caps lock support and show a warning if the user is entering an
SSH password with Caps Lock enabled
---------
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
Co-authored-by: 张小白 <364772080@qq.com>
Allows setting element as window control elements which consist of
`Drag`, `Close`, `Max`, or `Min`. This allows you to implement
dynamically sized elements that control the platform window, this is
used for areas such as the title bar. Currently only implemented for
Windows.
Release Notes:
- N/A
Fixes two things:
* ~3 months ago [in PR
#26420](https://github.com/zed-industries/zed/pull/26420/files#diff-33b58aa2da03d791c2c4761af6012851b7400e348922d64babe5fd48ac2a8e60)
`bindings_for_action` was changed to return bindings even when they are
shadowed (when the keystrokes would actually do something else).
* For edit prediction keybindings there was some odd behavior where
bindings for `edit_prediction_conflict` were taking precedence over
bindings for `edit_prediction` even when the `edit_prediction_conflict`
predicate didn't match. The workaround for this was #24812. The way it
worked was:
- List all bindings for the action
- For each binding, get the highest precedence binding with the same
input sequence
- If the highest precedence binding has the same action, include this
binding. This was the bug - this meant that if a binding in the keymap
has the same keystrokes and action it can incorrectly take display
precedence even if its context predicate does not pass.
- Fix is to check that the highest precedence binding is a full match.
To do this efficiently, it's based on an index within the keymap
bindings.
Also adds `highest_precedence_binding_*` variants which avoid the
inefficiency of building lists of bindings just to use the last.
Release Notes:
- Fixed display of keybindings to skip bindings that are shadowed by a
binding that uses the same keystrokes.
- Fixed display of `editor::AcceptEditPrediction` bindings to use the
normal precedence that prioritizes user bindings.
Fix a segfault in CVDisplayLink
We see 1-2 crashes a day on macOS on the `CVDisplayLink` thread.
```
Segmentation fault: 11 on thread 9325960 (CVDisplayLink)
CoreVideo CVHWTime::reset()
CoreVideo CVXTime::reset()
CoreVideo CVDisplayLink::runIOThread()
libsystem_pthread.dylib _pthread_start
libsystem_pthread.dylib thread_start
```
With the help of the Zed AI, I dove into the crash report, which looks
like this:
```
Crashed Thread: 49 CVDisplayLink
Exception Type: EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: KERN_INVALID_ADDRESS at 0x00000000000001f6
Exception Codes: 0x0000000000000001, 0x00000000000001f6
Thread 49 Crashed:: CVDisplayLink
0 CoreVideo 0x18c1ed994 CVHWTime::reset() + 64
1 CoreVideo 0x18c1ee474 CVXTime::reset() + 52
2 CoreVideo 0x18c1ee198 CVDisplayLink::runIOThread() + 176
3 libsystem_pthread.dylib 0x18285ac0c _pthread_start + 136
4 libsystem_pthread.dylib 0x182855b80 thread_start + 8
Thread 49 crashed with ARM Thread State (64-bit):
x0: 0x0000000000000000 x1: 0x000000018c206e08 x2: 0x0000002c00001513 x3: 0x0001d4630002a433
x4: 0x00000e2100000000 x5: 0x0001d46300000000 x6: 0x000000000000002c x7: 0x0000000000000000
x8: 0x000000000000002e x9: 0x000000004d555458 x10: 0x0000000000000000 x11: 0x0000000000000000
x12: 0x0000000000000000 x13: 0x0000000000000000 x14: 0x0000000000000000 x15: 0x0000000000000000
x16: 0x0000000182856a9c x17: 0x00000001f19bc540 x18: 0x0000000000000000 x19: 0x0000600003c56ed8
x20: 0x000000000002a433 x21: 0x0000000000000000 x22: 0x0000000000000000 x23: 0x0000000000000000
x24: 0x0000000000000000 x25: 0x0000000000000000 x26: 0x0000000000000000 x27: 0x0000000000000000
x28: 0x0000000000000000 fp: 0x000000016b02ade0 lr: 0x000000018c1ed984
sp: 0x000000016b02adc0 pc: 0x000000018c1ed994 cpsr: 0x80001000
far: 0x00000000000001f6 esr: 0x92000006 (Data Abort) byte read Translation fault
Binary Images:
0x1828c9000 - 0x182e07fff com.apple.CoreFoundation (6.9) <df489a59-b4f6-32b8-9bb4-9b832960aa52> /System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation
```
Using lldb to disassemble `CVHWTime::reset()` (and the AI to interpret
it), the crash is caused by dereferencing the pointer at the start of
the CVHWTime struct + 0x1c8. In this case the pointer has (the clearly
nonsense) value 0x2e (and 0x2e + 0x1c8 = 0x1f6, the failing address).
As to how this could happen...
Looking at the implementation of `CVDisplayLinkRelease`, it calls
straight into `CFRelease` on the main thread; and so it is not safe to
call `CVDisplayLinkRelease` concurrently with other threads that access
the CVDisplayLink. While we already stopped the display link, it turns
out that `CVDisplayLinkStop` just sets a flag on the struct to instruct
the io-thread to exit "soon", and returns immediately. That means we
don't know when the other thread will actually exit, and so we can't
safely call `CVDisplayLinkRelease`.
So, for now, we just leak these objects. They should be created
relatively infrequently (when the app is foregrounded/backgrounded), so
I don't think this is a huge problem.
Release Notes:
- Fix a rare crash on macOS when putting the app in the background.
Closes#16527
Release Notes:
- Added MacOS titlebar double-click action
---
Unfortunately, Apple doesn't seem to make the "Fill" API public or
documented anywhere.
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
Release Notes:
- N/A
----
The before version GPUI used `Cancel` for cancel text, if we use
non-English text (e.g.: "取消" in Chinese), then the press `Esc` to cancel
will not work.
So this PR to change it by use `PromptButton` to instead the `&str`,
then we can use `PromptButton::cancel("取消")` for the `Cancel` button.
Run `cargo run -p gpui --example window` to test.
---
Platform Test:
- [x] macOS
- [x] Windows
- [x] Linux (x11 and Wayland)
---------
Co-authored-by: Mikayla Maki <mikayla@zed.dev>
Co-authored-by: Mikayla Maki <mikayla.c.maki@gmail.com>
https://github.com/zed-industries/zed/issues/30972 brought up another
case where our context is not enough to track the actual source of the
issue: we get a general top-level error without inner error.
The reason for this was `.ok_or_else(|| anyhow!("failed to read HEAD
SHA"))?; ` on the top level.
The PR finally reworks the way we use anyhow to reduce such issues (or
at least make it simpler to bubble them up later in a fix).
On top of that, uses a few more anyhow methods for better readability.
* `.ok_or_else(|| anyhow!("..."))`, `map_err` and other similar error
conversion/option reporting cases are replaced with `context` and
`with_context` calls
* in addition to that, various `anyhow!("failed to do ...")` are
stripped with `.context("Doing ...")` messages instead to remove the
parasitic `failed to` text
* `anyhow::ensure!` is used instead of `if ... { return Err(...); }`
calls
* `anyhow::bail!` is used instead of `return Err(anyhow!(...));`
Release Notes:
- N/A
This glyphs field is usually larger than 8 elements, and SmallVec is not
efficient when it cannot store the value inline.
This change also adds precise glyphs run preallocation in some places
`ShapedRun` is constructed.
Release Notes:
- N/A
This PR adds a new `PlatformKeyboardLayout` trait with two methods:
`id(&self) -> &str` and `name(&self) -> &str`. The `id()` method returns
a unique identifier for the keyboard layout, while `name()` provides a
human-readable name. This distinction is especially important on
Windows, where the `id` and `name` can be quite different. For example,
the French layout has an `id` of `0000040C`, which is not
human-readable, whereas the `name` would simply be `French`. Currently,
the existing `keyboard_layout()` method returns what's essentially the
same as `id()` in this new design.
This PR implements the `name()` method for both Windows and macOS. On
Linux, for now, `name()` still returns the same value as `id()`.
Release Notes:
- N/A
Lukas Wirth
Gaauwe Rombouts
张小白
Floyd Wang
tidely
Piotr Osiewicz
Anthony Eid
Smit Barmase
Igal Tabachnik
Victor TranandMikayla Maki
Kirill Bulatov
Max Brunsfeld
Ridan VandenberghandMikayla Maki
Joseph T. Lyons
SunliandJason Lee
Hilmar WiegandandMichael Sloan
alphaArgonandPeter Tripp
Matin Aniss
Chung Wei LeongandMikayla Maki
Finn Evers
CharlesChen0823
