Now annotations are resolved with following algorithm:
1. On COMPILER_REQUIRED_ANNOTATIONS we resolve all annotations
and store results if this is compiler annotation, plugin annotation,
or annotation with meta-annotation (meta annotations are checked
recursively with designated resolution if needed)
2. On TYPES stage we resolve all those annotations once again and if
some annotation changes resolution then we keep type from p.1 and
report error on this annotation, so user should disambiguate it
Ambiguity may occur because of nested annotations with same name as
plugin annotations:
```
annotation class SomeAnnotation // (1) plugin annotation
open class Base {
annotation class SomeAnnotation // (2)
}
class Derived : Base() {
@SomeAnnotation // <-----------------
class Inner
}
```
At COMPILER_REQUIRED_ANNOTATIONS annotation call will be resolved to (1)
because at this stage supertypes are not resolved yet, and we consider
only importing scopes. At the TYPES stage we will find correct
annotation from supertype
Now predicates are split into LookupPredicate and DeclarationPredicate
hierarchies. First one allows to perform global search for declarations
and second one allows to check if some declaration matches the predicate.
Predicates with meta annotations are excluded from LookupPredicates,
because it's impossible to create index of annotations with meta-annotations,
because they can be located inside binary dependencies (so to achieve
this we need to scan the whole classpath).
Also only one predicate with meta-annotations is left in DeclarationPredicate
hierarchy (AnnotatedWithMeta)
^KT-53874 Fixed
^KT-53590 Fixed
This service allows resolve custom user types from plugins, which allows
to support specifying type arguments of generated supertypes basing
on arguments passed to annotations
Review: https://jetbrains.team/p/kt/reviews/6753
All redundant I managed to find, of course.
Why: I'm going to process all reflect dependencies in the next commits.
Cleanup reflect dependency before processing.
They are redundant because:
1. if `compileOnly` then compilation didn't break after dropping the
dependency
2. if `test*` then tests didn't break after dropping the dependency.
3. `analysis/analysis-api-fir/analysis-api-fir-generator/build.gradle.kts`
`compiler/fir/checkers/checkers-component-generator/build.gradle.kts`
Drop `implementation(project(":kotlin-reflect-api"))` because the
module already depends on
`implementation(project(":kotlin-reflect"))`
4. `compiler/daemon/daemon-client/build.gradle.kts`. Drop `runtimeOnly`
because after dropping `compileOnly` compilation didn't break (so
`runtimeOnly` looks suspicious). Less safe than 1-3
Original `ComponentRegistrar` exposes Project to its registration method,
so plugins should manually register extensions to it. To prepare for
possible unbound compiler from Project API in future new `K2ComponentRegistrar`
introduced which provides registration method without Project at all
This is needed to avoid dependency on :compiler:fir:fir2ir module in
backend parts of compiler plugins. It will allow to publish those
parts into jars for IDE, where they are needed for working of debugger
and bytecode toolwindow
- `ParentAnnotatedWith` matches declarations, which parent is annotated
- `HasAnnotatedWith` matches declarations, which have at least one
direct child with annotation
Also, `DeclarationPredicate.Any` is removed, because there is no
intention to support lookup for all declarations in module
^KT-52486 Fixed
There was a problem, that after founding of new annotation with meta
annotation we did not come back to previous files, which may contain
declarations with this new annotation
> FirPluginDiagnosticTestGenerated$MemberGen.testClassWithCompanionObject
> java.lang.IllegalStateException: dist/kotlin-stdlib-jvm-minimal-for-test.jar does not exist. Run 'gradlew dist'
Most of those tests are ignored, because IC is not supported properly
for compiler plugins. This commit is needed to prepare infrastructure,
fix current behaviour and collect main problems of IC with plugins
Check fail.txt files in new testdata for explanation of problems
Dmitriy Novozhilov