Commit graph

503 commits

Author SHA1 Message Date
Antonio Scandurra
0b2452f666
Merge pull request #1265 from zed-industries/worktree-performance
Fix problems that arise when large numbers of files change on disk
2022-07-04 09:19:18 +02:00
Antonio Scandurra
aca757a02d Don't poll snapshot if processing events unless user manually changed fs 2022-07-04 08:10:31 +02:00
Antonio Scandurra
cf05738f68 Remove redundant calls to poll_snapshot 2022-07-01 14:40:39 +02:00
Antonio Scandurra
a42399bcf3 Grab latest snapshot when invoking LocalWorktree::poll_snapshot 2022-07-01 14:36:51 +02:00
Antonio Scandurra
833aa726d5 🎨 2022-07-01 14:11:21 +02:00
Keith Simmons
ebe733a393 Restart language server using original root path rather than the path of the buffer restarted from 2022-06-30 19:11:21 -07:00
Max Brunsfeld
8a105bf12f WIP - try representing snapshots_to_send as a watch 2022-06-30 18:04:31 -07:00
Keith Simmons
832cc3dd19 Clear language server id for all worktrees when stopping a language server 2022-06-30 17:50:30 -07:00
Keith Simmons
6b50dda28a Fix failing test due to change in stop_language_server function 2022-06-30 17:40:50 -07:00
Keith Simmons
38ca4aab31 add assertion to test_definition ensuring no new language servers are created 2022-06-30 17:24:32 -07:00
Keith Simmons
37b75132b7 Minor comment change 2022-06-30 17:18:28 -07:00
Keith Simmons
f495185a4e add a comment explaining the three language server collections on project 2022-06-30 17:09:23 -07:00
Keith Simmons
5e7651e92e Kill starting servers as well as currently running ones 2022-06-30 17:00:29 -07:00
Keith Simmons
db05e32389 Prevent creating extra language server instances if there already exists one for that workspace 2022-06-30 16:46:26 -07:00
Max Brunsfeld
5fdbc38f46 Don't update worktrees' snapshots in the middle of processing fs events 2022-06-30 15:46:31 -07:00
Max Brunsfeld
b81135e10b Stop waiting for snapshot updates when disconnected from host 2022-06-30 15:07:40 -07:00
Antonio Scandurra
4ee8ee5a06 Ensure newer snapshots are always detected in wait_for_snapshot 2022-06-30 18:04:19 +02:00
Antonio Scandurra
484af8c7c4 Split worktree updates when a peer joins an already-shared project 2022-06-30 16:49:56 +02:00
Antonio Scandurra
845c79ee05 Respond to join project request before sharing project completes
This ensures the guest doesn't observe a huge delay when joining.
2022-06-30 14:29:06 +02:00
Antonio Scandurra
09bb3ddeb8 Split worktree updates and only send 256 entries at a time 2022-06-30 14:06:41 +02:00
Antonio Scandurra
5df0a6a425 Coalesce as many fs events as possible before processing them 2022-06-30 10:20:46 +02:00
Antonio Scandurra
b96962005e Keep looking for a newer snapshot before broadcasting it 2022-06-30 09:54:14 +02:00
Max Brunsfeld
8fe6809932 Exclude hidden worktrees from project's diagnostic summaries 2022-06-28 14:23:24 -07:00
Max Brunsfeld
f05e94d0de Allow diagnostics to be published for hidden worktrees 2022-06-28 13:31:04 -07:00
Antonio Scandurra
ca1d0a6e59 Ignore tokens that were not created via WorkDoneProgressCreate
With the new version of rust-analyzer, we were seeing stray `WorkDoneProgress::End`
messages that create an imbalance in the `pending_diagnostic_updates` that never
resolves. This was causing the diagnostic status bar item to never update because
we wouldn't emit `DiskBasedDiagnosticsStarted` nor `DiskBasedDiagnosticsFinished`.

This commit fixes the above situation by only acknowledging progress report for tokens
that have explicitly been created via the `WorkDoneProgressCreate` request, as stated
by the protocol.

In addition to that, we are replacing the `pending_diagnostic_updates: isize` with
a `has_pending_diagnostic_updates: bool`. We added it at some point to prevent a similar
issue where we would observe begin/end reports in a seemingly random order, which would cause
us to permanently display a `checking...` message in the status bar. I believe this commit
fixes that as well because the `isize` was just a less general solution for the same
underlying issue. As the protocol states: "the token provided in the create request should
only be used once (e.g. only one begin, many report and one end notification should be sent
to it)."
2022-06-28 10:08:43 +02:00
Keith Simmons
a477733bcb Address hover panic and add hint to install go when language server failed to install 2022-06-27 15:48:15 -07:00
Keith Simmons
bc82d98ae5
Merge pull request #1237 from zed-industries/jump-to-definition
Mouse jump to definition
2022-06-27 15:20:07 -07:00
Antonio Scandurra
7bfd7093b1 Don't stop scanning directory if reading one of the children errors 2022-06-27 08:40:37 +02:00
Keith Simmons
92ab107fc9 working jump to definition with tests 2022-06-24 15:09:31 -07:00
Keith Simmons
848445455d Working underline based on symbol origin 2022-06-24 15:05:35 -07:00
Keith Simmons
755636d10e Use NonZeroU32 to ensure settings tabsize cannot be zero 2022-06-23 12:55:38 -07:00
Antonio Scandurra
2cb8a3ccfb Don't override top-level settings with language defaults
Co-Authored-By: Nathan Sobo <nathan@zed.dev>
2022-06-23 16:40:57 +02:00
Antonio Scandurra
2d8ffbdfa2 Don't wait for host's worktree updates if they disconnected
Co-Authored-By: Max Brunsfeld <max@zed.dev>
2022-06-22 19:45:55 +02:00
Antonio Scandurra
2f09a4af16 Update project metadata when releasing a worktree's weak handle
When a project was getting unshared, we would remove invisible worktrees
from the project's state without notifying the server, which was causing
the randomized tests to fail.

This commit calls `Project::remove_worktree` when a worktree's handle gets
released. Doing so ensures that we update both our local metadata database
as well as the state on the server.
2022-06-22 10:26:10 +02:00
Antonio Scandurra
69aa3d848e Report running language servers when sharing project for the first time
Co-Authored-By: Max Brunsfeld <max@zed.dev>
2022-06-21 20:39:54 +02:00
Antonio Scandurra
4f215a77cd Don't compute fingerprint for every text summary 2022-06-17 15:15:27 +02:00
Antonio Scandurra
6f7a893ec9 Determine Buffer::is_dirty based on the rope's fingerprint 2022-06-17 12:38:25 +02:00
Antonio Scandurra
dc75b70724 Use word boundaries instead of syntax to infer completion edit ranges 2022-06-16 11:27:22 +02:00
Antonio Scandurra
35889add40 Add failing test for wrong typescript completion 2022-06-16 11:26:35 +02:00
Antonio Scandurra
3a1d0dd692 Track active projects in metrics
An active project is defined as a project where there has been at
least a buffer edit, a join request/response, or a follow update
in the last minute.
2022-06-15 10:33:20 +02:00
Keith Simmons
93158bfcff Fix delayed lsp request 2022-06-09 17:03:45 -07:00
Max Brunsfeld
87ba68e3ea
Merge pull request #1155 from zed-industries/golang
Add Go support
2022-06-09 14:18:37 -07:00
Max Brunsfeld
4ce4c0ef03 Ignore completions from gopls that we can't yet handle
We only support additionalEdits if they are provided when resolving the
completion, not if they are provided immediately.
2022-06-09 13:08:08 -07:00
Antonio Scandurra
213b31607c Remove language server statuses synchronously when stopping a server 2022-06-09 10:59:02 +02:00
Antonio Scandurra
69170fc33a Add unit test to ensure changing enable_language_server works 2022-06-09 10:48:06 +02:00
Antonio Scandurra
36a1a7a819 Start/stop language servers when enable_language_server changes 2022-06-09 10:08:11 +02:00
Antonio Scandurra
9d7476afc6 Extract a Project::stop_language_server method when restarting server
This will be useful later to stop a language server when detecting
configuration changes.
2022-06-09 10:05:35 +02:00
Max Brunsfeld
36ce3eb5ef Handle messages associated with LSP WorkDoneProgress::Begin messages
Co-authored-by: Antonio Scandurra <me@as-cii.com>
2022-06-08 10:26:41 -07:00
Antonio Scandurra
eda569d6b2 Snapshot file instead of path when creating a buffer snapshot 2022-06-08 14:29:14 +02:00
Antonio Scandurra
d8b885e3ec Provide tab_size when formatting document using LSP range formatting 2022-06-08 09:56:11 +02:00