* Add support for having channels in SimpleWorker
* Extract text model syncing code to a separate file
* Use a channel to do text model syncing for the language detection worker
* Simplify code
* Remove unused method
* Move OutputLinkComputer worker off editor worker
* Move TextMateTokenizationWorker off editor worker
* Simplify code
* Adopt channels for the host object
* More adopting channels for the host object
* More adopting channels for the host object
* More adopting channels for the host object
* Remove host object support from SimpleWorker
* Use the IEditorWorkerService, avoid starting a separate worker
* Bring the amd module id, the esm location and the worker label in a single type
* Improve typings
* SImplify worker creation pattern
* Enforce that all proxied methods start with `$` or `on`
* Adopt native proxy support
* Simplify code
* Simplify code
* Reintroduce a different standalone / workbench editor worker service
* Fix some always true/false errors in codebase
Testing out latest TS nightly that can identify when an expression is always true/false. This caught a few likely coding mistakes in our codebase
Unsure about the intent in some of these cases but I've tried my best to understand them. Pinging relevant code owners for the confusing cases
* Bump TS version
* Fix unreachable in build scripts
* Bump versions
* restore intent
* Revert package.json change
* Fixing a few new layer checker errors
These are global in node now
* Skip lib check in integration project
* One more skiplibcheck
* More skip lib
---------
Co-authored-by: Benjamin Pasero <benjamin.pasero@gmail.com>
* Make space for tree sitter
* Add the tree sitter wasm file
* Very naive tree-sitter syntax highlighting for html, with a layer breaker
* Update tree when content changes
* WIP for making abstract tokens class
* Handle theme changes
* Replace entire text model value with parse callback
* Perf improvements
* Add tree-sitter-typescript
* Add typescript + better initial parsing
* Refactor into tree parsing service and fix flaw in parse callback
* Remove things that aren't the parser service
* Add yielding
* Remove changes that aren't required for PR
* Remove more file changes
* Reduce yield to 50 ms
* Fix incremental parsing
* Try update node-abi
* Revert "Try update node-abi"
This reverts commit df28801e31.
* Update text buffer chunk api
* fix build
* Remove tree-sitter dependency
* Adopt new, as yet unpublished, `@vscode/tree-sitter-wasm` package
* Use published `@vscode/tree-sitter-wasm` package
* Break `TreeSitterTree` and `TreeSitterParserService` into better pieces
and:
- document the order of editor changes
- use service injection where `TextModel` is constructed
* Fix tests
* Remove unneeded import
* Fix missing tree-sitter-wasm in web and remote
* Make package.jsons match
* Add @vscode/tree-sitter-wasm to web loader config
* Try using importAMDNodeModule
* PR feedback
* Add race condition test for changing language while loading language
* Use same timeout
* Queue content changes
* Remove override dispose
* Move queue into TreeSitterTree
---------
Co-authored-by: Peng Lyu <penn.lv@gmail.com>
* Make space for tree sitter
* Add the tree sitter wasm file
* Very naive tree-sitter syntax highlighting for html, with a layer breaker
* Update tree when content changes
* WIP for making abstract tokens class
* Handle theme changes
* Replace entire text model value with parse callback
* Perf improvements
* Add tree-sitter-typescript
* Add typescript + better initial parsing
* Refactor into tree parsing service and fix flaw in parse callback
* Remove things that aren't the parser service
* Add yielding
* Remove changes that aren't required for PR
* Remove more file changes
* Reduce yield to 50 ms
* Fix incremental parsing
* Try update node-abi
* Revert "Try update node-abi"
This reverts commit df28801e31.
* Update text buffer chunk api
* fix build
* Remove tree-sitter dependency
* Adopt new, as yet unpublished, `@vscode/tree-sitter-wasm` package
* Use published `@vscode/tree-sitter-wasm` package
* Break `TreeSitterTree` and `TreeSitterParserService` into better pieces
and:
- document the order of editor changes
- use service injection where `TextModel` is constructed
* Fix tests
* Remove unneeded import
* Fix missing tree-sitter-wasm in web and remote
* Make package.jsons match
* Add @vscode/tree-sitter-wasm to web loader config
* Try using importAMDNodeModule
* PR feedback
* Add race condition test for changing language while loading language
* Use same timeout
* Queue content changes
* Remove override dispose
* Move queue into TreeSitterTree
---------
Co-authored-by: Peng Lyu <penn.lv@gmail.com>
* add integration tests for LanguageModelChat#sendRequest
* make sure errors are recreated when making LM requests
* disable test with a note for later
* fix remote integration tests
* don't dimiss inline chat content widget when already having typed something
https://github.com/microsoft/vscode-copilot/issues/6067
* extract `TextOnlyMenuEntryActionViewItem` for reuse
* remove unused variables
* * show chat input below request/response pairs
* setting for text-only buttons
* more dynamic buttons
* always show the first request, don't repopulate input with last message
* keep progress bar hidden, rely on "Generating..."
* no more special background color
* add `minimal` renderer style for chat renderings
* tweak font-size for details when render mode is minimal
* stable scroll position for inline chat, don't push down the lines chat is editing but push the inline chat upwards
* more buttons more compact, tweak labels
* * add missing service dependency
* repopulate input for some unit test
* allow output from `InteractiveChatController` suite
Bug (previous behavior):
1.Press Ctrl+Shift+F to open the search panel.
2.Enter any text into the search input.
3.Press PageUp/PageDown.
4.The layout shifts to the right, hiding the line that indicates the current tab in the activity bar.
Solution:
When search results are displayed, pressing PageUp or PageDown now moves the cursor to the start or end of the search input, respectively. This is achieved by assigning specific functionality to these keys, overriding their default behavior of moving through the page.
Testing:
A smoke test was implemented to assess the fix. The test involves entering text into the search editor, pressing the PageUp and PageDown keys, and checking for any changes in the activity bar's layout.
* add `ListenerRefusalError` and `ListenerLeakError` which get logged when listener thresholds are exceeded.
The `stack` property of these errors will point towards the most frequent listener and how often it is used. If that's a high number there is a leak (same listener is added over and over again), if that's a low number there might be a conceptual flaw that an emitter is simply too prominent.
* rightfully don't use Error.captureStackTrace (v8/nodejs only)
Alexandru Dima
Benjamin Pasero
Connor Peet
Robo
Peng Lyu
Alex Ross
Peng Lyu
Megan Rogge
Daniel Imms
Matt Bierner
Ulugbek Abdullaev
Johannes Rieken
Francisca Carneiro
Benjamin Christopher Simmonds
Logan Ramos