I recently noticed that on my Windows 11 machine, Zed no longer receive
the `WM_DWMCOLORIZATIONCOLORCHANGED` message when the system theme
changes. This functionality was present in the past. While this change
might be unexpected, it's understandable given Microsoft's history of
system updates.
This pull request proposes an alternative approach using the
`WM_SETTINGCHANGE` message to handle theme changes.
Release Notes:
- N/A
Fix bugs caused by the window context PR, where the window could be on
the stack and is then requested from the App.
This PR also adds derive macros for `AppContext` and `VisualContext` so
that it's easy to define further contexts in API code, such as
`editor::BlockContext`.
Release Notes:
- N/A
Fix a bug where a GPUI macro still used `ModelContext`
Rename `AsyncAppContext` -> `AsyncApp`
Rename update_model, read_model, insert_model, and reserve_model to
update_entity, read_entity, insert_entity, and reserve_entity
Release Notes:
- N/A
Closes#23621
Change was in #23378. Also adds a comment to clarify why this is
inconsistent with all other uses of `bindings_for_action`.
Release Notes:
- N/A
There's still a bit more work to do on this, but this PR is compiling
(with warnings) after eliminating the key types. When the tasks below
are complete, this will be the new narrative for GPUI:
- `Entity<T>` - This replaces `View<T>`/`Model<T>`. It represents a unit
of state, and if `T` implements `Render`, then `Entity<T>` implements
`Element`.
- `&mut App` This replaces `AppContext` and represents the app.
- `&mut Context<T>` This replaces `ModelContext` and derefs to `App`. It
is provided by the framework when updating an entity.
- `&mut Window` Broken out of `&mut WindowContext` which no longer
exists. Every method that once took `&mut WindowContext` now takes `&mut
Window, &mut App` and every method that took `&mut ViewContext<T>` now
takes `&mut Window, &mut Context<T>`
Not pictured here are the two other failed attempts. It's been quite a
month!
Tasks:
- [x] Remove `View`, `ViewContext`, `WindowContext` and thread through
`Window`
- [x] [@cole-miller @mikayla-maki] Redraw window when entities change
- [x] [@cole-miller @mikayla-maki] Get examples and Zed running
- [x] [@cole-miller @mikayla-maki] Fix Zed rendering
- [x] [@mikayla-maki] Fix todo! macros and comments
- [x] Fix a bug where the editor would not be redrawn because of view
caching
- [x] remove publicness window.notify() and replace with
`AppContext::notify`
- [x] remove `observe_new_window_models`, replace with
`observe_new_models` with an optional window
- [x] Fix a bug where the project panel would not be redrawn because of
the wrong refresh() call being used
- [x] Fix the tests
- [x] Fix warnings by eliminating `Window` params or using `_`
- [x] Fix conflicts
- [x] Simplify generic code where possible
- [x] Rename types
- [ ] Update docs
### issues post merge
- [x] Issues switching between normal and insert mode
- [x] Assistant re-rendering failure
- [x] Vim test failures
- [x] Mac build issue
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Cole Miller <cole@zed.dev>
Co-authored-by: Mikayla <mikayla@zed.dev>
Co-authored-by: Joseph <joseph@zed.dev>
Co-authored-by: max <max@zed.dev>
Co-authored-by: Michael Sloan <michael@zed.dev>
Co-authored-by: Mikayla Maki <mikaylamaki@Mikaylas-MacBook-Pro.local>
Co-authored-by: Mikayla <mikayla.c.maki@gmail.com>
Co-authored-by: joão <joao@zed.dev>
#23411 introduced an "Accept" callout for single line edits, but the
logic to detect them was incorrect causing it to trigger for multiline
insertions, this PR fixes that.
Release Notes:
- N/A
Closes#22653 again
In PR #23283, I thought that every `runnable` dispatched to the main
thread would correspond to an `EVENT_DISPATCHED` message in the message
queue. However, after testing, some `runnable`s occasionally weren’t
executed.
This PR updated the code as follows:
```rust
if let Ok(runnable) = self.main_receiver.try_recv() { <-- before
for runnable in self.main_receiver.drain() { <-- after
runnable.run();
}
```
This ensures that runnables are handled more proactively on the main
thread, now we handle `runnable`s with a much higher priority.
A big thanks to @MolotovCherry and @ArthurBrussee for their testing
efforts!
Release Notes:
- N/A
Considered doing this when previously working on completions menu
layout, as it brings the default selection position next to the cursor
position, and is generally more symmetrical. With #23445 there is now a
more compelling reason, as the "translucent, cropped bottom" display
doesn't make sense when displayed above.
Release Notes:
- N/A
Closes#23015
Release Notes:
- Improved which keybindings are selected for display. Now later entries
within `bindings` will take precedence. The default keymaps have been
updated accordingly.
Adds support for Cut, Copy, Paste, Undo, Redo, New, Open, Save, and Find
keys to the default keymap. These keys can be found on old keyboards,
but also custom layouts like
[Extend](https://dreymar.colemak.org/layers-extend.html).
Release Notes:
- Added support for the Cut, Copy, Paste, Undo, Redo, New, Open, Save,
and Find keys to the default keymap.
This PR renames the constants and functions previously introduced in
PR#23283. Since the changes are within the GPUI crate, I renamed these
from `**_ZED_**` to `**_GPUI_**`.
Release Notes:
- N/A
* Collects and reports all parse errors
* Shares parsed `KeyBindingContextPredicate` among the actions.
* Updates gpui keybinding and action parsing to return structured
errors.
* Renames "block" to "section" to match the docs, as types like
`KeymapSection` are shown in `json-language-server` hovers.
* Removes wrapping of `context` and `use_key_equivalents` fields so that
`json-language-server` auto-inserts `""` and `false` instead of `null`.
* Updates `add_to_cx` to take `&self`, so that the user keymap doesn't
get unnecessarily cloned.
In retrospect I wish I'd just switched to using TreeSitter to do the
parsing and provide proper diagnostics. This is tracked in #23333
Release Notes:
- Improved handling of errors within the user keymap file. Parse errors
within context, keystrokes, or actions no longer prevent loading the key
bindings that do parse.
It's easy to overshoot the bottom of the tooltip when cursoring to a
button, such as opening the commit from a blame tooltip. Before this
change the tooltip would immediately disappear, and now it sticks around
for a bit.
Also:
* Shares the implementation with `elements/text.rs`. This will
particularly be handy when it makes use of hoverable tooltips.
* Improves the fix to #21657.
- Now the element will no longer think it has an active tooltip that it
registers with the window.
- It will instead display the next available tooltip, whereas I believe
before the next available tooltip would be suppressed.
* Fixes bug where `cx.refresh()` wasn't called when text tooltip is
hidden due to a mouse down event.
* Ports over fix in https://github.com/zed-industries/zed/pull/14832 to
`elements/text.rs`
Release Notes:
- The tooltip for inline git blame now waits a bit before disappearing
when the mouse leaves it.
Closes#22653
After some investigation, I found this bug is due to that sometimes
`foreground_task` is not dispatched to the main thread unless there is
user input. The current Windows implementation works as follows: when
the `WindowsDispatcher` receives a `foreground_task`, it adds the task
to a queue and uses `SetEvent(dispatch_event)` to notify the main
thread.
The main thread then listens for notifications using
`MsgWaitForMultipleObjects(&[dispatch_event])`.
Essentially, this is a synchronous method, but it is not robust. For
example, if 100 `foreground_task`s are sent, `dispatch_event` should
theoretically be triggered 100 times, and
`MsgWaitForMultipleObjects(&[dispatch_event])` should receive 100
notifications, causing the main thread to execute all 100 tasks.
However, in practice, some `foreground_task`s may not get a chance to
execute due to certain reasons.
As shown in the attached video, when I don't move the mouse, there are
about 20-30 `foreground_task`s waiting in the queue to be executed. When
I move the mouse, `run_foreground_tasks()` is called, which processes
the tasks in the queue.
https://github.com/user-attachments/assets/83cd09ca-4b17-4a1f-9a2a-5d1569b23483
To address this, this PR adopts an approach similar to `winit`. In
`winit`, an invisible window is created for message passing. In this PR,
we use `PostThreadMessage` to directly send messages to the main thread.
With this implementation, when 100 `foreground_task`s are sent, the
`WindowsDispatcher` uses `PostThreadMessageW(thread_id,
RUNNABLE_DISPATCHED)` to notify the main thread. This approach enqueues
100 `RUNNABLE_DISPATCHED` messages in the main thread's message queue,
ensuring that each `foreground_task` is executed as expected. The main
thread continuously processes these messages, guaranteeing that all 100
tasks are executed.
Release Notes:
- N/A
Context pills are now focusable and intractable via the keyboard.
- <kbd>←</kbd> and <kbd>→</kbd> move the focus to the previous or next
item (wrapping if necessary)
- <kbd>↓</kbd> and <kbd>↑</kbd> move the focus vertically
- If the cursor is in the first/last row of the assistant/inline editor,
they will move the focus to the strip
- Inside the strip, they will move the focus to the pill horizontally
overlapping the most
- If already in the first/last row of the strip, they will move to the
first/last pill (like in editors)
- If the first/last pill is focused, they will move the focus back to
the editor
- <kbd>⌫</kbd> removes the focused pill (unless it's the suggested one)
- <kbd>⏎</kbd> accepts the suggested pill if focused
https://github.com/user-attachments/assets/040bc71c-a3ae-4961-9886-2d5c3d290a73
Release Notes:
- N/A
Various fixes for Zeta and one fix that's visible to non-Zeta-using
users of inline completions.
Release Notes:
- Changed inline completions (Copilot, Supermaven, ...) to not show up
in empty buffers.
---------
Co-authored-by: Antonio <antonio@zed.dev>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Bennet <bennet@zed.dev>
rust-analyzer does not support derive_macro expansion in attributes -
https://github.com/rust-lang/rust-analyzer/issues/8092. This could be
worked around via a proc_macro, but I think it'd be best to just require
docs for every action.
Release Notes:
- N/A
Also:
* Adds `impl_internal_actions!` for deriving the `Action` trait without
registering.
* Removes some deserializers that immediately fail in favor of
`#[serde(skip)]` on fields where they were used. This also omits them
from the schema.
Release Notes:
- Keymap settings file now has more JSON schema information to inform
`json-language-server` completions and info, particularly for actions
that take input.
No issue, as the functionality is currently not being used in Zed. This
is more of a GPUI improvement.
Currently, `keyboard_layout` and `on_keyboard_layout_change` are already
handled on macOS. This PR implements the same for X11 and Wayland.
Linux supports up to 4 keyboard layout groups (e.g., Group 0: English
US, Group 1: Bulgarian, etc). On X11 and Wayland, `event` provides a new
active group, which maps to the `layout_index`. We already store keymap
state from where we can get the current `layout_index`. By comparing
them, we determine if the layout has changed.
X11:
<img
src="https://github.com/user-attachments/assets/b528db77-1ff2-4f17-aac5-7654837edeb9"
alt="x11" width="300px" />
Wayland:
<img
src="https://github.com/user-attachments/assets/2b4e2a30-b0f4-495c-96bb-7bca41365d56"
alt="wayland" width="300px" />
Release Notes:
- N/A
Closes#22433
Before/After (macOS):
For some reason the non-blurred one seems much lower quality, so we may
need to tinker with the samples, or something else.
I'm unsure if this is a problem on Linux/in the Blade renderer, but
since no changes were made outside of the medal shaders we can probably
take this macOS-specific win for now.
Release Notes:
- gpui: Fixed an issue where shadows with a `blur_radius` of 0 would not
render.
- [x] Rewrite worktree git handling
- [x] Fix tests
- [x] Fix `test_propagate_statuses_for_repos_under_project`
- [x] Replace `WorkDirectoryEntry` with `WorkDirectory` in
`RepositoryEntry`
- [x] Add a worktree event for capturing git status changes
- [x] Confirm that the local repositories are correctly updating the new
WorkDirectory field
- [x] Implement the git statuses query as a join when pulling entries
out of worktree
- [x] Use this new join to implement the project panel and outline
panel.
- [x] Synchronize git statuses over the wire for collab and remote dev
(use the existing `worktree_repository_statuses` table, adjust as
needed)
- [x] Only send changed statuses to collab
Release Notes:
- N/A
---------
Co-authored-by: Cole Miller <cole@zed.dev>
Co-authored-by: Mikayla <mikayla@zed.com>
Co-authored-by: Nathan <nathan@zed.dev>
Closes#22326
This PR adds process PID information to window created by X11, so that
window manager can identify which process this window belongs to.
Without this property, the window manager would have no reliable way to
know which process created this window.
In original issue, `robotgo` throws error on `x, y, w, h :=
robotgo.GetBounds(pid)` this method. If we go deeper into the source
code of `robotgo`, it calls `GetXidFromPid` which goes through all
windows, and tries to check for provided pid. Hence, when it tries to do
that for Zed, it fails and returns `0, err` to caller.
```go
// Robotgo source code trying to look through all windows and query pid
// GetXidFromPid get the xid from pid
func GetXidFromPid(xu *xgbutil.XUtil, pid int) (xproto.Window, error) {
windows, err := ewmh.ClientListGet(xu)
if err != nil {
return 0, err
}
for _, window := range windows {
wmPid, err := ewmh.WmPidGet(xu, window)
if err != nil {
return 0, err
}
if uint(pid) == wmPid {
return window, nil
}
}
return 0, errors.New("failed to find a window with a matching pid.")
}
```
Querying for pid for active Zed window:
Before:
```sh
tims@lemon ~/w/go-repro [127]> xprop -root _NET_ACTIVE_WINDOW
_NET_ACTIVE_WINDOW(WINDOW): window id # 0x4e00002
tims@lemon ~/w/go-repro> xprop -id 0x4e00002 _NET_WM_PID
_NET_WM_PID: not found.
```
After:
```sh
tims@lemon ~/w/go-repro> xprop -root _NET_ACTIVE_WINDOW
_NET_ACTIVE_WINDOW(WINDOW): window id # 0x4e00002
tims@lemon ~/w/go-repro> xprop -id 0x4e00002 _NET_WM_PID
_NET_WM_PID(CARDINAL) = 103548
tims@lemon ~/w/go-repro>
```
Correct zed process PID (below) assosiated with zed window (shown
above):
Release Notes:
- Fix `robotgo` failing when Zed window is open on Linux
Closes#22578
Currently, the `hovered` boolean in the window state is only updated by
the `WM_MOUSELEAVE` event, which fires when the mouse cursor leaves the
window's working area. This means that when the user moves the cursor
from the window to the title bar, `hovered` is set to `false`. Later in
the code, this flag is used to determine the cursor style and check if
the cursor is over the correct window.
The `hovered` boolean should remain active even when the mouse is over
non-client items, such as the title bar or window borders. This PR fixes
that by using `WM_NCMOUSELEAVE` event, which is triggered when the mouse
leaves non-client items. This event is used to update the `hovered`
boolean accordingly.
Now, `hovered` is `true` when the mouse is over the window's working
area, as well as non-client areas like the title bar.
More context:
- Existing: `dwFlags: TME_LEAVE` tracks window area mouse leaves, which
is used in `handle_mouse_move_msg` func.
- New: `dwFlags: TME_LEAVE | TME_NONCLIENT` tracks non-client mouse
leaves, which is used in `handle_nc_mouse_move_msg` func.
Preview:
https://github.com/user-attachments/assets/b319303f-81b9-45cb-bf0c-535a59b96561
Release Notes:
- Fix cursor style not changing on hover over items in the title bar on
Windows
This commit is all about strings, not about line layout at all. When
laying out text, we use a line layout cache to avoid roundtrips to
system layout engine where possible. This makes it so that we might end
up not needing an owned version of text to insert into the cache, as we
might get a cached version.
The API boundary of line layout accepted text to be laid out as &str. It
then performed cache lookup (which didn't require having an owned
version) and only resorted to making an owned version when needed. As it
turned out though, exact cache hits are quite rare and we end up needing
owned version more often than not. The callers of line layout either
dealt with SharedStrings or owned Strings. Due to coercing them into
&str, we were ~always copying text into a new string (unless there was a
same-frame-hit). This is a bit wasteful, thus this PR generifies the API
a bit to make it easier to reuse existing string allocations if there
are any.
Benchmark scenario: scrolling down page-by-page through editor_tests (I
ran the same scenario twice):
Release Notes:
- N/A
Closes https://github.com/zed-industries/zed/issues/21657
Follow-up of https://github.com/zed-industries/zed/pull/22488
Previous PR broke git blame tooltips, which are expected to be open when
hovered, even if the mouse cursor is moved away from the actual blame
entry that caused the tooltip to appear.
Current version moves the invalidation logic into `prepaint_tooltip`,
where the new data about the tooltip origin is used to ensure we
invalidate only tooltips that have no mouse cursor in either origin
bounds or tooltip bounds (if it's hoverable).
Release Notes:
- Fixed tooltips getting stuck
This reverts commit 344284e013.
That change broke git blame tooltips, as Zed should also show tooltips
which are hovered, even though the mouse had left the origin element's
bounds.
Release Notes:
- N/A
Closes https://github.com/zed-industries/zed/issues/21657
In case of the task rerun button tooltip from
f6dabadaf7/crates/terminal_view/src/terminal_view.rs (L1051-L1070)
, the actual button element is not styled as invisible, only its parent.
Zed won't render such element since it's parent is hidden, but will
consider it "visible" all the time its `paint` is called, spawning a
task with the delay, that will create the tooltip:
f6dabadaf7/crates/gpui/src/elements/div.rs (L1949-L1959)
When the parent is hidden, the child won't be painted anymore, and no
mouse listeners will be able to detect this fact and hide the tooltip.
Hence, check such cases separately, during `prepaint`, and invalidate
the tooltips that are not valid anymore.
We cannot use `hitbox.is_hovered(cx)` as it's not really hovered during
prepaint, so a mouse position check is used instead.
Release Notes:
- Fixed tooltips getting stuck
* Remove unnecessary WindowContext and ViewContext '_ lifetimes
* Removed some cases where WindowContext has a different name than `cx`.
Release Notes:
- N/A
