* Lazily compute signature type predicates
* Use an instance of IdentifierTypePredicate to represent an unresolved type predicate
* Simplify `getMaybeTypePredicate`
* Invert representation of `resolvedTypePredicate`
* Remove `__unresolvedTypePredicate` type and remember to use `noTypePredicate` instead of `undefined` when in all `createSignature` calls
* Fix style of getTypePredicateOfSignature
* Use in createGetSymbolWalker
* Fix bugs for unions of type predicates
* Code review
* Make noTypePredicate purely an implementation detail of getTypePredictateOfSignature
* Add test
* Add test for #19642
* Add test with reversed order
* Still instantiate the extends clause even when theres a noimplicitany error in js
* Only be permissive for JS
* In JS, instantiate classes even when they have too many type arguments, instead of returning unknownType
* Error when LHS of instanceof is Union of Primitives #18519
* Refactor to allTypesAssignableToKind method and update test
* Use ternary expression instead of if / else blocks
* Do visibility painting from collectLinkedAliases in checker to remove statefullness in declaration emit
* Fix#17085
* Add deeply destructured array to test
* Add test case for #18634
* Add PR feedback
* Treat `...` in jsdoc type as creating a synthetic rest parameter -- not as an array type
* Change type parsing so `...T[]` parses as `...(T[])` and not `(...T)[]`
* Replace the last parameter with ...args, and make access to it potentially undefined
* Code review
* Update JSX intrinsic element test to match babel’s
* Update baseline
* Move regex out of isIntrinsicJsxName
* Switch to non-regex intrinsic name test
* Consider the commonjs module indicator as an indicator that something is effectively an external module
* Only use commonjs module indicator when targeting commonjs
so that error message refers to prop name instead of param name.
Changed getter errors for similar case so they also refer to prop name.
Fixed bug where static getters wouldn't output their specific error.
Fixes#1976
* Use documentation comments from inherited properties when @inheritDoc is present
The JSDoc `@ineheritDoc` [tag](http://usejsdoc.org/tags-inheritdoc.html)
"indicates that a symbol should inherit its documentation from its
parent class". In the case of a TypeScript file, this also includes
implemented interfaces and parent interfaces.
With this change, a class method or property (or an interface property)
with the `@inheritDoc` tag in its JSDoc comment will automatically use
the comments from its nearest ancestor that has no `@inheritDoc` tag.
To prevent breaking backwards compatibility,
`Symbol.getDocumentationComment` now accepts an optional `TypeChecker`
instance to support this feature.
fixes#8912
* Use ts.getJSDocTags as per @andy-ms 's recommendation
* Convert @inheritDoc tests to verify.quickInfoAt
* Concatenate inherited and local docs when @inheritDoc is present
* Make typeChecker param explicitly `TypeChecker | undefined`
* Re-accept baseline after switch to explicit `| undefined`
* Update APISample_jsodc.ts to match new getDocumentationComment signature
* Re-accept baselines after rebasing
* Instantiate contextual types while in an inferrential context
* Limit scope of instantiation to only when likely needed
* Still get aparent type
* Expand test
* Fix nit
* Handle JSX and array
* Tests for the JSX and Array cases
* After much deliberation and inspection, much simpler fix
After much deliberation and inspection, much simpler fix
Undo
Redo
* Revised emit for computed property names
* Fix downlevel name generation scopes
* Accept slightly more conservative baseline
* First feedback pass
* Reduce number of nonrequired variable declarations and assignments
* Remove side-effect-free identifier references
* skip partially emitted expressions
* Comments, move starsOnNewLine to emitNode
* Put expressions on newlines when inlined in class expressions for consistency
* Update new ref
* Fix typo in comment
