The panic happened when a user had a settings file with a buffer line
height custom variant, because the drop-down renderer only took into
account the two named variants.
The fix for this will be creating a custom element that allows a user to
manually input a line height greater than one or select either
Comfortable or Standard.
Release Notes:
- N/A
The font picker from onboarding is a lot friendlier to interact with and
makes it impossible for a user to select an invalid font from the
settings ui.
I also moved the font picker from the onboarding crate to the ui_input
crate
## New Look
<img width="1136" height="812" alt="image"
src="https://github.com/user-attachments/assets/7436682c-6a41-4860-a18b-13e15b8f3f31"
/>
Release Notes:
- N/A
mac_watcher already does this so it would make more sense to also do
this on Windows and it saves ~500-600mb of ram on the chromium project.
This does not improve memory usage on linux because inotify cannot do
recursive directory monitoring
Release Notes:
- N/A
## Summary
Fixes an issue where font features (like ligatures) were not applied to
text under the vim block cursor. The cursor would inherit the font
family from the character at the cursor
position, but would use default font features instead of the editor's
configured font features.
## Changes
- Make the font mutable when rendering the vim block cursor
- Apply the editor's text style font features to the cursor font
This ensures that text under the block cursor renders with the same
visual appearance as the rest of the editor content.
Closes#39471
Release Notes:
- Fixed vim block cursor not respecting font features (like ligatures)
Should close#39428
The working directory of the `wsl.exe` program is set to a Linux path,
which is invalid on the Windows side, causing the terminal to crash. The
first spawn works because there is no active terminal view, allowing a
new shell (which checks for the remote) to be created. I cannot explain
why it works on SSH remote clients, but I may be missing something in
the remote connection implementation.
I don't have a Windows machine to test this, so I would appreciate
someone testing it. 🙏🏼
Release Notes:
- Fixed an issue where WSL terminals could not be splitted
---------
Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>
This PR adds the numeric stepper component to the settings ui and
implements some settings that rely on this component as well.
I also switched {buffer/ui}_font_weight to the `gpui::FontWeight` type
and added a manual implementation of the Schemars trait. This allows Zed
to send min, max, and default information to the JSON LSP when a user is
manually editing the settings file.
The numeric stepper elements added to the settings ui are below:
- ui font size
- ui font weight
- Buffer font size
- Buffer font weight
- Scroll sensitivity
- Fast scroll sensitivity
- Vertical scroll margin
- Horizontal scroll margin
- Inline blame padding
- Inline blame delay
- Inline blame min column
- Unnecessary code fade
- Tab Size
- Hover popover delay
Release Notes:
- N/A
# Why
I have find out that this tool exists by browsing Keymap Editor. I think
it would be nice for its discoverability to show it in the app menus in
Dev builds.
# How
Add "GPUI Inspector" app menu item conditionally for Dev builds only.
Release Notes:
- N/A
# Preview
<img width="1014" height="948" alt="Screenshot 2025-10-01 at 14 36 48"
src="https://github.com/user-attachments/assets/c0409e67-1f4d-44f3-90b3-293ad4fe5c73"
/>
We have unnecessary clones for the fields here as most of the snapshots
contain the others hierarchically.
Release Notes:
- N/A *or* Added/Fixed/Improved ...
Add support for rendering Unicode combining characters (diacritics) in
the terminal's batched text runs.
- Add append_zero_width_chars() to handle combining marks
- Integrate zero-width chars into all batching code paths
- Update cell extras tracking logic
- Add test for combining character rendering
Fixes display of é, ñ, ô and other diacritics.
Closes#39525
Release Notes:
- Fixed: NFD/NFKD normalized text (e.g., é as e + ◌́) not rendering in
integrated terminal
Before:
<img width="874" height="688" alt="SCR-20251004-udnj"
src="https://github.com/user-attachments/assets/8d9f9c9f-dac4-4382-92c2-8b6c1d817abd"
/>
After:
<img width="873" height="686" alt="SCR-20251004-ulsw"
src="https://github.com/user-attachments/assets/fbd5cdc7-fdd6-44dc-8b05-cc425644f1a0"
/>
The editor settings control module was the first prototype of what a
settings UI could look like in Zed, but the code is outdated now and is
no longer used. So this PR removes it for cleanup.
Release Notes:
- N/A
I was looking at the rope implementation and some of the existing bugs
that crash in there, and I ran cargo-mutants to inspect test coverage. I
was motivated by bugs like
https://github.com/zed-industries/zed/issues/38556 but this doesn't fix
it and the bug may well be at a higher layer.
This PR adds coverage for a few functions that aren't tested today. I
didn't find any actual bugs yet.
I can see this tree is pretty sparse on docstrings so if you think these
are too verbose I can take them out or drop the whole PR.
Release Notes:
- N/A
When running flycheck, I've noticed that scrolling starts to lag:
https://github.com/user-attachments/assets/b0bef0a3-ccbd-479d-a385-273398086d38
When checking the trace, it is notable that project panel updates its
entire tree multiple times during flycheck:
<img width="2032" height="1136" alt="image"
src="https://github.com/user-attachments/assets/d1935e77-3b00-4be5-a12a-8a17a9d64202"
/>
[scrolling.trace.zip](https://github.com/user-attachments/files/22710852/scrolling.trace.zip)
Turns out, `target/debug` directory is loaded by Zed (presumably,
reported by langserver as there are sources generated by bindgen and
proto that need to be loaded), and `target/debug/build` directory
received multiple events of a `None` kind for Zed, which trigger the
rescans.
Rework the logic to omit the `None`-kind events in Zed, and to avoid
excessive repo updates if not needed.
Release Notes:
- Improved worktree FS event emits in gitignored directories
---------
Co-authored-by: Cole Miller <cole@zed.dev>
Follow-up: https://github.com/zed-industries/zed/pull/38590
**Note**: this PR contains changes from the [previous
PR](https://github.com/zed-industries/zed/pull/38590), when that PR gets
merged we should see the real changes.
This PR fixes 4 things in order to make:
1. Add html/markdown minifier to remove all the **\t** and **\n**
characters. This is needed as you cannot create new lines with markdown
by just adding an enter to the source file.
2. The event Event::HTML only contained a chunk of the real html for
multiline HTML code. I fixed this by storing the currently watched HTML
inside a buffer and at the end we parse it into the right elements.
Instead of trying to parse a chunck into multiple elements which would
always fail before.
3. Add support for html tables.
4. Fixed panic that occured when table does not have an header.
I also decided to keep the html minifier inside Zed, because making it a
dependency for just a few 100 lines seems to be an overkill. The
original crate had a few cve in their dependencies, so figured this
would be the best.
**Html table support**
<img width="1439" height="801" alt="Screenshot 2025-09-27 at 12 19 07"
src="https://github.com/user-attachments/assets/a884cc6f-cf47-45a2-81fa-91300c7bbf3f"
/>
**Before & after Zed's README (no changes)**
<img width="3440" height="1378" alt="Screenshot 2025-09-27 at 12 34 47"
src="https://github.com/user-attachments/assets/1273b094-fb24-4abd-bffa-56ef3b44670c"
/>
Release Notes:
- Markdown: Added support for html tables
## Problem
When splitting a terminal pane, the new pane opens in the root directory
(`/`) instead of preserving the current working directory of the
original terminal.
For example, when working in `/Users/modestnerd/Developer/Projects/zed`
(my pc) and splitting the terminal pane, the new pane would open in `/`
instead of staying in the current directory.
## Solution
Restructured the fallback logic in
`new_pane_with_cloned_active_terminal` (terminal_panel.rs:452-456) to
ensure `default_working_directory(workspace, cx)` is called as a
fallback even when a terminal view exists but its `working_directory()`
returns `None`.
The fix changes the nested `and_then` to use `or_else` for the fallback,
ensuring the working directory is always properly resolved before
entering the async block.
Release Notes:
- Fixed terminal split pane opening in wrong directory instead of
preserving the current working directory
