* --explainFiles currently hardcoded
* Move configFileSpecs to configFile so it can be used in program later
* Explain root file inclusion reason and explain include files in the log
* Baseline explainFiles
* Fix incorrectly reporting of file list two times in --b mode
* Fix unnecessary new lines in output represented incorretly in the baseline
* More tests
* More cleaning up
* Keep listing files in same order as list files, just add explaination
* Fix double listing of file names when the program has errors
* Make diagnostic chains for file include reason
* Add explaination for the file include to diagnostics for program
* Harness ls incorrectly adding tsconfig as the root file
* Fix incorrect use of path for calculating absolute path
* Fix the root file in fourslash
* Test project service options merge
* Add config file name to matched by include explaination
* Add test for when the file changes and program is reused completely but related file information is reattached to correct location
* Handle file preprocessing diagnostics updates when program is reused and related information location changes
* Moved types to types.ts
* Refactoring and cleanup
* More cleanup
* More refatoring
* Handle synthetic imports
* Baselines after merge
* Adds experimental support for running TS Server in a web worker
This change makes it possible to run a syntax old TS server in a webworker. This is will let serverless versions of VS Code web run the TypeScript extension with minimal changes.
As the diff on `server.ts` is difficult to parse, here's an overview of the changes:
- Introduce the concept of a `Runtime`. Valid values are `Node` and `Web`.
- Move calls to `require` into the functions that use these modules
- Wrap existing server logic into `startNodeServer`
- Introduce web server with `startWebServer`. This uses a `WorkerSession`
- Add a custom version of `ts.sys` for web
- Have the worker server start when it is passed an array of arguments in a message
In order to make the server logic more clear, this change also tries to reduce the reliance on closures and better group function declarations vs the server spawning logic.
**Next Steps**
I'd like someone from the TS team to help get these changes into a shippable state. This will involve:
- Adddress todo comments
- Code cleanup
- Make sure these changes do not regress node servers
- Determine if we should add a new `tsserver.web.js` file instead of having the web worker logic all live in `tsserver.js`
* Shim out directoryExists
* Add some regions
* Remove some inlined note types
Use import types instead
* Use switch case for runtime
* Review and updates
* Enable loading std library d.ts files
This implements enough of `ServerHost` that we can load the standard d.ts files using synchronous XMLHttpRequests.
I also had to patch some code in `editorServices`. I don't know if these changes are correct and need someone on the TS team to review
* Update src/tsserver/webServer.ts
* Update src/tsserver/webServer.ts
Co-authored-by: Sheetal Nandi <shkamat@microsoft.com>
* Addressing feedback
* Allow passing in explicit executingFilePath
This is required for cases where `self.location` does not point to the directory where all the typings are stored
* Adding logging support
* Do not create auto import provider in partial semantic mode
* Handle lib files by doing path mapping instead
* TODO
* Add log message
This replaces the console based logger with a logger that post log messages back to the host. VS Code will write these messages to its output window
* Move code around so that exported functions are set on namespace
* Log response
* Map the paths back to https:
// TODO: is this really needed or can vscode take care of this
How do we handle when opening lib.d.ts as response to goto def in open files
* If files are not open dont schedule open file project ensure
* Should also check if there are no external projects before skipping scheduling
Fixes failing tests
* Revert "Map the paths back to https:"
This reverts commit 0edf650622.
* Revert "TODO"
This reverts commit 04a4fe7556.
* Revert "Should also check if there are no external projects before skipping scheduling"
This reverts commit 7e4939014a.
* Refactoring so we can test the changes out
* Feedback
Co-authored-by: Sheetal Nandi <shkamat@microsoft.com>
* Loosen author tag parsing
Also make the code more efficient and easier to read.
1. Arbitrary text, except for newline and @, is allowed for the author name.
2. Arbirtrary text, except for newline, is allowed for the email
address.
3. Newline is treated as a match for an open <
I tried to allow newlines in the author and email, but it was ambiguous
with the end of the tag.
I also got rid of the two lookaheads and unified the trailing comment
handling.
Fixes#41804
* remove hardmode test since it did not reveal anything new
* Baseline showing #41801 and other issues with output path calculation
* Add a way to note descripencies between clean and incremental build
* Add descripency when no rootDir is specified but project is composite
* if rootDir is specified, irrespective of whether all files belong to rootDir, the paths should be calculated from rootDir
* Fix the output file names api to use the correct common source directory
* Tests for #41780
* Spelling
* Less aggressive wildcard check, 'keyof any' constraint for 'infer T' in mapped type constraint position
* Accept new baselines
* Add regression tests
* Allow type-only ImportEqualsDeclarations
* Suppress CJS-in-ESM error when type-only
* Add grammar error on import type in import alias
* Update API baselines
* Fix importsNotUsedAsValues with ImportEqualsDeclarations
* Make bad error talk words more good for Daniel. Fixes#41603
* One more error message baseline update
* Update transformer and emitter
