* Retain comments on (and produce sourcemaps on) the equals token in initializers
* Improved comments/sourcemaps for await, yield, while, and for
* Retain comments on block curly-braces
* Emit comments for if statements
* Improved switch case comment emit
* Improve comment and sourcemap emit for try/catch, throw, and continue statements
* Improve sourcemap emit and comments for with statements
* More accurate sourcemaps+comments emit for new, typeof, void, and delete
* Improve comment emit for element access expressions
* Preserve more comments on imports and exports
* Make function a bit more defensive like other usages of emitTrailingCommentsOfPosition
* Support preserving comments within empty lists
* Handle leading comments of tokens, conditionally indent leading comments
* Stop heuristically sourcemapping tokens
When the transform was trivial it worked, but was unneeded, but when it was complex, it was brittle - best leave source mapping up to the transformers
* Fix unneeded +1
* Tighten up element access comments
* Handle comments on parenthesized expression tokens
* Fix nit
* Have getNameOfSymbolAsWritten quote nonidentifier nonnumeric symbols all the time
* Revert checker changes
* Reuse synthetic origin to indicate that derived declaration name may need to be quoted
* Use a limited version of getApparentType that doesnt map primitives
* Reuse [most of] getBaseConstraintOfType, since it does the needed behaviors
* Move new function next to the very similar function
* Changed "Duplicate Identifier" to "enum can only be merged..."
when either declaration of the identifier is an enum.
Partial (?) fix for #529
Not sure if the new test is necessary, all the cases seem to have been covered by others tests.
* picked a nit
* Use getAllowSyntheticDefaultImports to access `--allowSyntheticDefaultImport` value
* Fix#21788: Handel missing imporotClause case
* Fix#21789: Add a defensive check to forgottenThisPropertyAccess code fix for non-identifier locations
* Do not suggest prefix with `this` if the name we are looking for is diffrent from the errorLocation
* Fix#21796: Handel case of unknown module
* Add check to capture more info for #21800
* Fix#21807: check for symbol before looking up its flags
* Fix#21812: Gracefully fail if the token is not `this`.
* fix: `matchedText` was not properly extract the correct parts from the candidate
closes: #21636
* core: replace `substr` with `substring` in `matchedText`
* test: remove redundant code from tests
* Fix JSX contextual types to not eagerly become apparent
* Apply changes from code review, unify common code
* Fix jsx children contextual typing
* Light code review feedback
* Use fillMissingTypeArguments
* Accept nonliteral jsx child type
* Add test for the fillMissingTypeArguments case
* Only apply global augmentations before globals are available
* Add detailed comment explaining the split of global/nonglobal augmentations
* Remove trailing whitespace
* Overloads in Array.concat now handle ReadonlyArray
Previously it was union types, which is slower.
* Make arrayConcat3 test stricter
* Switch to InputArray instead of adding overloads
* Update baselines
* Update baselines correctly
* Rename to ConcatArray and add slice method
Should make it, respectively, easier to understand this specific type
and harder to satisfy it by mistake.
Turns out SemicolonClassElement is a specific kind for semicolons inside
a class. Having one of them with --noUnusedLocals on would crash the
compiler after the assert added in #21013.
Changes are in src/compiler.checker.ts only
The second arguments to the function "removeOptionalityFromDeclaredType" has been changed from "getRootDeclaration(declaration)" to "declaration".
* Use apparent type of original type to handle indexes
* Redo older fix causing new bug by extending getDeclaredOrApparentType instead of getTypeWithFacts
* Rename symbol
