Before this change we had a race condition bug: if stderr was closed
before the other two sockets, we wouldn't properly detect when the
server died, and not report or retry anything.
That's because we treated a closed stderr as a non-error.
Technically, it isn't an error (closing a connection is okay!), but
until we have a proper shutdown ceremony between all three processes, we
can treat it as an error, because that lets us to detect when the server
is gone.
On the client-side, we also always react to these errors by
reconnecting. Except when we shutdown: there we do a proper shutdown and
won't error on the proxy exit code.
So, this works, even if I wish there was a better way for the server to
communicate to the proxy that it shutdown properly. But I don't want a
fourth socket.
Release Notes:
- N/A
* uses the state that's synced, to fetch the language server name
* uses proper, canonicalized path when creating a remote ssh worktree,
otherwise `~/foo/something` stays unexpanded
Release Notes:
- N/A
This PR makes it so collab will trigger a refresh for a user's LLM token
whenever their subscription changes.
This allows us to proactively push down changes to their subscription.
In order to facilitate this, the Stripe event processing has been moved
from the `api` service to the `collab` service in order to access the
RPC server.
Release Notes:
- N/A
This restores the change from #19193 that I erroneously reverted in
#19234.
I think the bug in #19275 got in my way when testing.
With that bug fixed, the changes in here also work fine.
Release Notes:
- N/A
I've been a bit annoyed by the hover divider rule being extremely bright
compared to other divider rules in the UI. This PR updates their color
to use the regular border color from the current theme instead of the
muted (but still pretty bright) text color.
Apologies for the unsolicited PR (and please feel free to close if it
goes against some other plans / designs you already have in place :).
#### Example screenshot before:
<img width="302" alt="Screenshot 2024-10-15 at 23 29 18"
src="https://github.com/user-attachments/assets/7ea22808-8135-4a46-9457-e670225aebaa">
#### Example screenshot after:
<img width="312" alt="Screenshot 2024-10-15 at 23 28 16"
src="https://github.com/user-attachments/assets/63ac0d02-ae6d-4962-84a2-1fdb95519b15">
***
Release Notes:
- Make the divider rule in LSP hovers more muted
Release Notes:
- Added a new `editor: format selections` action that allows formatting
only the currently selected text via the primary language server.
---------
Co-authored-by: Thorsten Ball <mrnugget@gmail.com>
When the SSH command dies or the server, the channel gets dropped and
the heartbeat method went into an infinite loop causing a hang.
Oversight from yesterday. Fixed now.
Release Notes:
- N/A
Closes#19040
Addresses the problem with annoying error messages on windows (see
comment from SomeoneToIgnore on #18567)
Release Notes:
- Fixed the bug where language servers from PATH would sometimes be
prioritised over the ones from `direnv`
- Stopped running environment loading on windows as it didn't work
anyways due to `SHELL` not being set
This commit adds support for using the physical insert-button. First
click toggles insert mode and subsequent clicks toggle back and forth
between replace and insert mode.
Closes#19224
Release Notes:
- Added support for using the insert button for vim_mode.
Part of https://github.com/zed-industries/zed/issues/19239
Overall, this hardcoding approach has to stop and Zed better show some
notification/modal that proposes to select a primary language server,
when launching with the language that has no such settings.
Release Notes:
- Fixed Astro LSP interactions
Supersedes https://github.com/zed-industries/zed/pull/19166
TODO:
- [x] Update basic zed paths
- [x] update create_state_directory
- [x] Use this with `NodeRuntime`
- [x] Add server settings
- [x] Add an 'open server settings command'
- [x] Make sure it all works
Release Notes:
- Updated the actions `zed::OpenLocalSettings` and `zed::OpenLocalTasks`
to `zed::OpenProjectSettings` and `zed::OpenProjectTasks`.
---------
Co-authored-by: Conrad <conrad@zed.dev>
Co-authored-by: Richard <richard@zed.dev>
This PR refines the SSH modal UI, adjusting spacing and alignment. Via
these changes, I'm also introducing the ability for the `empty_message`
on the `List` component to receive not just a string but any element.
The custom way in which the SSH modal was designed made it feel like
this was needed for proper spacing.
<img width="700" alt="Screenshot 2024-10-16 at 1 20 54 AM"
src="https://github.com/user-attachments/assets/f2e0586b-4c9f-4497-b4cb-e90c8157512b">
Release Notes:
- N/A
This PR makes it so the Storybook loads GPUI with the default features
enabled.
This fixes a panic that would occur when trying to run any of the
stories.
Release Notes:
- N/A
Sometimes, issues are created outside of issue templates (which we don't
prefer, but we can't prevent). This updates our top-ranking issues
script such that it will add `triage` and `admin read` labels to any
issue that is missing a core label, so that we don't miss the issues
when doing the next triage.
Release Notes:
- N/A
when no result found, always display `Search all files`, which is
confused.
Release Notes:
- Fixed an issue where the project search would sometimes show "Search
all files" when there were no results.
---------
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
This does two things.
Important one: it reverts #19193, which lead to our whole process
handling breaking. When the `proxy` process was killed, it apparently
didn't close the stdout/stderr anymore, which meant we would not detect
when it died. (Watching its `status()` in the io loop also didn't work!)
We should figure out how to keep our process handling working before we
make this change in #19193, which sounds reasonable.
Second, less important thing: I think we should treat the process being
killed from a signal as non-zero, as an error.
Release Notes:
- N/A
This improves the heartbeat detection logic. We now treat any other
incoming message from the ssh remote server
as a heartbeat message, meaning that we can detect re-connects earlier.
It also changes the connection handling to await futures detached.
Co-Authored-by: Thorsten <thorsten@zed.dev>
Release Notes:
- N/A
---------
Co-authored-by: Thorsten <thorsten@zed.dev>
Co-authored-by: Antonio <antonio@zed.dev>
TL;DR: Another O(n^2) strikes.
In #19194 we received a report about a 7Mb JSON file that Zed struggles
with. Naturally this file showcased a O(n^2) in line layout; this file
has one long line.
During line layout for Mac we have to convert between UTF-16 and UTF-8
indices in the string, as CoreText works with UTF-16 and Rust strings
are UTF-8. The problem stemmed from the fact that we were re-seeking our
string converter on each glyph, which boils down to: we were reparsing
[0..curr_string_position] bytes up to full length of the string, which
is the O(n^2) in question. This PR changes this behaviour to reuse the
Index Converter if the position we're seeking to is not yet reached.
Basically, we're treating the converter as forward iterator and we try
to seek with the same iterator, if possible.
Where previously you could not even open the file in OP (within
reasonable time frame, I waited for 40 seconds before giving up), now
you can do it in.. slightly over a second. The best part is: the
experience is still not ideal. Typing in the buffer is sluggish. Still,
this is a start.
Release Notes:
- Mac: Improved performance with very long lines
This PR changes the SSH modal design so its more keyboard
navigation-friendly and adds the server nickname feature.
Release Notes:
- N/A
---------
Co-authored-by: Danilo <danilo@zed.dev>
Co-authored-by: Danilo Leal <67129314+danilo-leal@users.noreply.github.com>
This also puts the entire state of `StripeBilling` behind a `RwLock`.
When fetching the existing prices and meters, or when inserting new
ones, we acquire a write lock and hold it until the Stripe request
completes. This prevents two concurrent calls to `get_or_insert_price`
from inserting the same data twice.
Creating a new meter/price is unusual, so in practice we'll acquire a
read lock most of the time.
/cc @rtfeldman @maxdeviant
Release Notes:
- N/A
Closes#19051Closes#19182
#### How to reproduce this crash:
1. Open any file and input some ASCII characters.
2. Replace these characters with `你好`.
3. Press `backspace`.
4. Crash.
https://github.com/user-attachments/assets/ea5c5340-29a5-42c8-98c5-6e60770445a4
The issue lies with the `prefix_offset` introduced in #18858. After the
buffer is modified, this value is not always valid and may fall within a
`char boundary`, which results in a crash.
Release Notes:
- Fixed Supermaven crashing on deleting non-ASCII text
remote: Disable ControlPersist for master ssh connection
`ControlPersist=yes` combined with `ControlMaster=yes` silently forces
`ForkAfterAuthentication=yes` (even when the user has explicitly set it
to `no` - reported upstream in [0]) - and the latter makes the ssh
subprocess disappear, which makes us think that the connection died
(This is only an issue for people who have `ControlPersist=yes` in their
`ssh_config`, and perhaps the answer is "if that option breaks things,
don't use that option?" - but it's an option that makes sense _most_ of
the time, it's just in this edge-case of "creating an ssh connection
with -N and expecting the process to stay in the foreground" where it
_must_ be set to no)
I think the alternative approach is to tell people "if you want to use
persistent connections, have a separate ~/.ssh/config entry for
servername (to ssh into) and servername-no-persist (to zed into)", which
is possible, but ugh. Kind of a messy situation >.<
Tests:
- Before: Connections to my server result in "Failed to connect: ." (The
error message is attempting to show stderr, but stderr is empty)
- After: Connections to my server work reliably
[0] https://bugzilla.mindrot.org/show_bug.cgi?id=3743
Release Notes:
- N/A
