Walking up the tree to find the enclosing SourceFile would distort the
timing too much so, instead, we attach a Path in the binder.
At present, the path is determined retroactively by walking up the call
stack in the trace visualizer, but this is both inconvenient and
routinely inaccurate (checking an expression in one file may require
checking an expression in another file and there's no way to determine
from the trace where this transition occurred).
* Initial support for module: node12
* Add allowJs and declaration emit enabled tests
* Fix typos
* cts, mts, cjs, mjs, etc extension support
* Fix watch of files whose intepretation changes due to a package.json update
* Minor PR feedback
* Adjust error message
* Initial import/export/self-name support
* Accept new error codes
* TypesVersions support in export/import map conditions
* Fix import suggestion and autoimport default extensions under new resolution modes
* Add tests for import maps non-relative name lookup feature
* Fix isDeclarationFileName for .d.mts and .d.cts
* Preserve new extensions when generating module specifiers
* Fix spurious implict any suggestion caused by file ordering bug and optimize import name format detection by relying on parents being set
* Fix a bunch of incremental bugs that dont repro under fourslash for some reason
* Accept updated baseline
* Always include extensions on completions for cjs/mjs style imports
* String completion relative import suggestions respect the mode of the import when choosing if they provide extensions
* Style feedback
* Change diagnostic case
* fix for 45006
* treat setters like getters in preceding commit; move test accordingly
* fix test baselines
* changes per code review
in `getContainerFlags`, move cases for get/set accessors
to fallthrough into the block that currently handles MethodDeclaration;
so get/set accessors and method declarations all get the same container flags,
such that during `bindContainer`, `startFlow.node` is assigned to
getters/accessors
(this changes a public api in tsserverlibrary.d.ts and typescript.d.ts
by adding `GetAccessorDeclaration` and `SetAccessorDeclaration` to the type
of `FlowStart.node`)
consolidate predicates checking whether a node is either a get or set
accessor, into `isObjectLiteralOrClassExpressionMethodOrAccessor`
(formerly `isObjectLiteralOrClassExpressionMethod`)
annotate updated test with `@target: es2020`
* fix `isObjectLiteralOrClassExpressionMethodOrAccessor`
require that Getter/Setters are parented by an ObjectLiteralExpression or ClassExpression
* Visit children of jsdoc type aliases in the binder
This sets up parent pointers.
Fixes#45254 and almost certainly #45248, though I haven't figured out
to repro the second case.
* move incorrect parenthesis
* manually set comment parent instead
* Bind children of typedef where possible
* add explanatory comment to binding
* Class static block (#9)
* Add types factory and parser
* Add some case
* Make class static block as a container
* Update cases
* Add visitor
* Add emitter and more compile target
* Check boundary of break and continue
* Add basic transformer
* Fix emit behavior
* Add more tests
* Add friend tests
* Update baseline
* Fix cr issues
* Accept baseline
* Add decorator and modifier check
* Add functional boundary check
* Fix conflict
* Fix computed prop name within context
* Add more tests
* Update baseline
* Avoid invalid test baseline
* Support use before initialize check
* wip
* Fix class static block context
* Fix checks
* Fix missing case
* Improve assert message
* Accept baseline
* Avoid new context
* Update diagnostic message
* Fix name collision
* Fix targets
* Avoid unnecessary files
* Add more case
* Add more test cases
* Fix strict mode function declaration
* Avoid private fields initializer if no private identifier references
* Avoid private fields and add more test case
* Add more case
* Add tests and support for related services functionality
* Fix this reference in static block
* Split parser diagnostic and binder diagnostic
Co-authored-by: Ron Buckton <ron.buckton@microsoft.com>
* Cache parsed path mapping patterns
If a project has many of them (e.g. 1800), parsing the patterns
repeatedly can take up a lot of time.
* Move cache to ConfigFileSpecs
* Inline constants
* Simplify cache access
* support QualifiedName when narrowing inside loops
* add test
* narrow more qualified names
* handle `undefined` of `getFlowCacheKey `
* update comments in test
* Improve errors for incorrectly nested export default
The compiler and services don't handle incorrectly nested
`export default` well right now:
```ts
export = (x,y) => {
export default { }
}
```
Asking for document highlights, find all references or quick info on
'export' or 'default' cause a crash. After the crash is fixed, the error
message is confusing and wrong: "An export assignment cannot be used outside a module."
This PR:
1. Skips document highlights for incorrectly nested export default.
2. Skips find all refs for incorrectly nested export default.
3. Switches the fallback binding for incorrectly nested export default
from Alias to Property. Neither is correct, but Property doesn't cause a
crash in alias resolution.
4. Improves the error message to reflect a post-ES module world, which
has export default and 'module' means 'ES module', not 'namespace'.
Fixes#40082 and the related bugs mentioned above.
* address PR comments
JSDoc typedefs don't actually have hosts, because they're not
semantically attached to a declaration. However, the parser still
attaches them to some declaration (or statement), but that declaration
is not related to the typedef.
Previously, delayedBindJSDocTypedefTag used getJSDocHost to walk past
the unrelated declaration, but #41858 correctly started categorising
typedefs as unattached, with no host, so the binder began falling
back to file scope.
The path to skip the unrelated declaration is always the same, though, so this
PR uses `typeAlias.parent.parent` instead of `getJSDocHost(typeAlias)`.
* Allow redeclaring CommonJS alias with initial undefined
This allows us to read our own output, plus the times when people
manually write exactly the same pattern.
Fixes#40555
* Use control flow to type commonjs exports
1. Could probably use a *lot* more tests.
2. getTypeOfAlias redoes some work from resolveAlias because it needs to
not resolve the alias completely, just to its export.
* fix lint, improve jsdoc
* Add tests, improve+fix control flow
1. Update the module.exports test to match the exports ones.
2. Add a test of evolving commonjs type.
3. Add a test of assignment as last statement.
(1) exposed a bug that required a better synthetic reference. (3)
exposed a bug that was most easily fixed by giving source files a
`endFlowNode` like functions and setting it in the binder.
* fix lint
* About halfway through the checker
I'm going to merge with master to avoid clashing with the declaration
fix.
* Add undefined to Symbol.valueDeclaration
Also add undefined to a number of utility functions that have always
accepted it, but never added it to their type.
* Fix lint from code review
Co-authored-by: Daniel Rosenwasser <DanielRosenwasser@users.noreply.github.com>
* remove obsoleted fix from inferFromUsage
Co-authored-by: Daniel Rosenwasser <DanielRosenwasser@users.noreply.github.com>
* Add undefined to Symbol.declarations' type
Symbol.declarations now has type `Declaration[] | undefined`.
I made a mistake somewhere in the checker related to JS checking, so
there are quite a few test failures right now.
* undo clever change to getDeclaringConstructor
* Address PR comments
1. More early-returns.
2. More line breaks.
* Refactor emit substitution into transform
* Add reusable state machine for binary expressions
* Allow emitBinary to use state machine for comments/sourcemaps
* Switch core trampoline state back to arrays
* Switch binder to parallel stacks, temporarily partially revert emitBinary
* Add link to benchmark when posting perf results
* Ensure work stacks are per-execution
* Reenable comments and sourcemaps
* Eliminate well-known symbols in the checker: 2021 edition
* Actually update the lib text to say unique symbol, too (this is unneeded with compat code in place, but this makes goto-def make more sense)
* Add test showing mismatched symbol constructor type interop
* Add more test cases for some other related issues this fixes
* Revert computed name change
* Style comments
Make `tracing` either `undefined` or the same namespace as before.
Switching all calls to `tracing?.___` means that there is no cost for
a call or the arguments when tracing is not used. Comparing two runs
without tracing (27 runs, drop 5+5, avg rest) I get:
master:
42.59s user 1.00s system 165% cpu 26.372 total
changed:
42.01s user 0.982 system 165% cpu 26.039 total
(Makes it all private, so no api changes.)
* chore: failing test for const enums and isolatedModules
* fix: const enums + isolatedModules emit invalid code
In `isolatedModules` mode, the compiler does not inline const enums,
but also decides not to `import` them, leaving invalid code that
throws a `ReferenceError` at runtime.
This code:
```
import { SomeEnum } from './bar';
sink(SomeEnum.VALUE);
```
..should compile to either:
```
var { SomeEnum } = require('./bar');
sink(SomeEnum.VALUE);
```
..or (with const enum inlining):
```
sink(1 /* VALUE */);
```
..but actually compiles to:
```
sink(SomeEnum.VALUE);
```
..with no imports, which throws a ReferenceError at runtime.
---
The compiler has already realised that the symbol is a referenced const
enum, it just doesn't use this information when it comes to deciding
whether to emit an import. This commit additionally checks that
information, if we are compiling in isolatedModules mode.
---
In my opinion, this is not the correct fix, but it is the simplest. In
`isolatedModules` mode, `const enum` should probably be a compile error
(because there are no benefits and the complexity is high, and,
apparently, buggy). If not, the compiler should not be checking whether
something is a const enum, and should just be treating it as a regular
enum, and checking as if it was?
Fixes#40499.
* chore: extra test for type-only
* feat: explicitly ban --isolatedModules --preserveConstEnums false
* feat: isolatedModules implies preserveConstEnum
* Update src/compiler/diagnosticMessages.json
Co-authored-by: Andrew Branch <andrewbranch@users.noreply.github.com>
* chore: compiler test for argument incompatibility
* Add and fix test for namespace import of const enum
* Fix additional bug with reexport of const-enum-only module
Co-authored-by: Andrew Branch <andrewbranch@users.noreply.github.com>
Co-authored-by: Andrew Branch <andrew@wheream.io>
Storing the arguments on the stack will make it possible to forego
try-finally blocks when we start tracing in server scenarios, which have
to handle cancellation.
* Fix: `E` events need to have the same information that is on the
corresponding `B` events. (Superseded below.)
* Use `I` (not `i`) for instant events, so they show in devtools
too. (Though they don't go through the flame chart as they do in
`about://tracing`, so they're not nearly as useful.)
* Abstract the code that writes the records in a single `writeEvent`
local function.
* Make `args` optional, and default to `undefined` (which will not add
them) at all.
* Drop the `{ "ts": ... }` wrapper around the `args`, after verifying
that having arguments with names like `begin`, `end`, `pos`, `id`
doesn't interfere with either UIs.
* Add `tracing.push`/`tracing.pop` for complete events, change a few
`.begin`/`.end` to use these. (The caveat is that until there's an exit
handler to dump unterminated events, these won't show in the dump. When
that's done the push/pop variant can be used everywhere.)
* Add meta lines to name the process and the thread, and a line that
avoids the warning when opening in devtools.
* Fix error on duplicate commonjs exports
Previously, the code missed setting the parent pointer for the lhs
access expression.
Also add declaration emit of element access expressions, missed in my
previous PR.
* Switch to excludes=None, add test case
CommonJS exports have None excludes, but still have an error issued by
the checker. This is the previous behaviour even though it would be nice
to add some exclusions.
