Release Notes:
- Fixed horizontal scrolling not working for sticky items in the Project
Panel.
- Fixed issue where hovering over the last sticky item in the Project
Panel showed a hovered state on the entry behind it.
- Improved behavior when clicking a sticky item in the Project Panel so
it scrolls just enough for the item to no longer be sticky.
This solves problems where users couldn't shut down sessions while
locators or build tasks are running.
I renamed `debugger::Session::Mode` enum to `SessionState` to be more
clear when it's referenced in other crates. I also embedded the boot
task that is created in `SessionState::Building` variant. This allows
sessions to shut down all created threads in their boot process in a
clean and idiomatic way.
Finally, I added a method on terminal that allows killing the active
task.
Release Notes:
- Debugger: Allow shutting down debug sessions while they're booting up
Follow-up #33868
This PR fixes a few issues with determining the completion range for
client‑ and variable‑list completions.
1. Non‑word completions
We previously supported only word characters and _, using their combined
length to compute the start offset. In PHP, however, an expression can
contain `$`, `-`, `>`, `[`, `]`, `(`, and `)`. Because these characters
weren’t treated as word characters, the start offset stopped at them,
even when the preceding character was part of a word.
2. Trailing characters inside the search text
When autocompletion occurred in the middle of the search text, we didn’t
account for trailing characters. As a result, the start offset was off
by the number of characters after the cursor. For example, replacing res
with result in print(res) produced `print(rresult)` because the trailing
`)` wasn’t subtracted from the start offset.
The following completions are correctly covered now:
- **Before** `$aut` -> `$aut$author` **After** `$aut` -> `$author`
- **Before** `$author->na` -> `$author->na$author->name` **After**
`$author->na` -> `$author->name`
- **Before** `$author->books[` -> `$author->books[$author->books[0]`
**After** `$author->books[` -> `$author->books[0]`
- **Before** `print(res)` -> `print(rresult)` **After** `print(res)` ->
`print(result)`
**Before**
https://github.com/user-attachments/assets/b530cf31-8d4d-45e6-9650-18574f14314chttps://github.com/user-attachments/assets/52475b7b-2bf2-4749-98ec-0dc933fcc364
**After**
https://github.com/user-attachments/assets/c065701b-31c9-4e0a-b584-d1daffe3a38chttps://github.com/user-attachments/assets/455ebb3e-632e-4a57-aea8-d214d2992c06
Release Notes:
- Debugger: Fixed autocompletion not always replacing the correct search
text
This PR updates the debugger panel's session list to be more useful in
some cases that are commonly hit when using the JavaScript adapter. We
make two adjustments, which only apply to JavaScript sessions:
- For a child session that's the only child of a root session, we
collapse it with its parent. This imitates what VS Code does in the
"call stack" view for JavaScript sessions.
- When a session has exactly one thread, we label the session with that
thread's name, instead of the session label provided by the DAP. VS Code
also makes this adjustment, which surfaces more useful information when
working with browser sessions.
Closes#33072
Release Notes:
- debugger: Improved the appearance of JavaScript sessions in the debug
panel's session list.
---------
Co-authored-by: Julia <julia@zed.dev>
Co-authored-by: Remco Smits <djsmits12@gmail.com>
Closes#34323Closes#34313
The previous PR #33932 introduced a way to "close" the
`pending_requests` buffer of the `TransportDelegate`, preventing any
more requests from being added. This prevents pending requests from
accumulating without ever being drained during the shutdown sequence;
without it, some of our tests hang at this point (due to using a
single-threaded executor).
The bug occurred because we were closing `pending_requests` whenever we
detected the server side of the transport shut down, and this closed
state stuck around and interfered with the retry logic for SSH+TCP
adapter connections.
This PR fixes the bug by only closing `pending_requests` on session
shutdown, and adds a regression test covering the SSH retry logic.
Release Notes:
- debugger: Fixed a bug causing SSH connections to some adapters
(Python, Go, JavaScript) to fail and restart endlessly.
Closes#34029
The crash is due to a stack overflow in our `html_to_markdown`
conversion; I've added a maximum depth of 200 for the recursion in that
crate to guard against this kind of thing.
Separately, we were treating all content-types other than `text/plain`
and `application/json` as HTML; I've changed this to only treat
`text/html` and `application/xhtml+xml` as HTML, and fall back to
plaintext. (In the original crash, the content-type was
`application/octet-stream`.)
Release Notes:
- agent: Fixed a potential crash when fetching large non-HTML files.
Closes#33342
This PR implements serialization of pinned tabs regardless of their
state (empty, untitled, etc.)
The root cause was that empty untitled tabs were being skipped during
serialization but their pinned state was still being persisted, leading
to a mismatch between the stored pinned count and actual restorable
tabs, this issue lead to a crash which was patched by @JosephTLyons, but
this PR aims to be a proper fix.
**Note**: I'm still evaluating the best approach for this fix. Currently
exploring whether it's necessary to store the pinned state in the
database schema or if there's a simpler solution that doesn't require
schema changes.
---
**Edit from Joseph**
We ended up going with altering our recall logic, where we always
restore all editors, even those that are new, empty, and unsaved. This
prevents the crash that #33335 patched because we are no longer skipping
the restoration of pinned editors that have no text and haven't been
saved, throwing off the count dealing with the number of pinned items.
This solution is rather simple, but I think it's fine. We simply just
restore everything the same, no conditional dropping of anything. This
is also consistent with VS Code, which also restores all editors,
regardless of whether or not a new, unsaved buffers have content or not.
https://github.com/zed-industries/zed/tree/alt-solution-for-%2333342
Release Notes:
- N/A
---------
Co-authored-by: Joseph T. Lyons <JosephTLyons@gmail.com>
Closes #ISSUE
An idea I and @MrSubidubi came up with, to improve UX around the
keystroke input.
Currently, there's a hard tradeoff with what to focus first in the edit
keybind modal, if we focus the keystroke input, it makes keybind
modification very easy, however, if you don't want to edit a keybind,
you must use the mouse to escape the keystroke input before editing
something else - breaking keyboard navigation.
The idea in this PR is to have a dual-phased focus system for the
keystroke input. There is an outer focus that has some sort of visual
indicator to communicate it is focused (currently a border). While the
outer focus region is focused, keystrokes are not intercepted. Then
there is a keybind (currently hardcoded to `enter`) to enter the inner
focus where keystrokes are intercepted, and which must be exited using
the mouse. When the inner focus region is focused, there is a visual
indicator for the fact it is "recording" (currently a hacked together
red pulsing recording icon)
<details><summary>Video</summary>
https://github.com/user-attachments/assets/490538d0-f092-4df1-a53a-a47d7efe157b
</details>
Release Notes:
- N/A *or* Added/Fixed/Improved ...
Closes #ISSUE
Makes it so that `KeymapFile::update_keybinding` treats removals of
bindings that weren't user-defined as creating a new binding to
`zed::NoAction`.
Release Notes:
- N/A *or* Added/Fixed/Improved ...
Now DAP logs show the label of each session which makes it much easier
to pick out the right one.
Also "initialization sequence" now shows up correctly when that view is
selected.
Release Notes:
- N/A
---------
Co-authored-by: Cole Miller <cole@zed.dev>
