This is the way it works in K1.
See `OverloadingConflictResolver.kt:345`.
^KT-57568 Fixed
Merge-request: KT-MR-9395
Merged-by: Nikolay Lunyak <Nikolay.Lunyak@jetbrains.com>
Why did the problem existed?
At first, lambdas were analyzed with suspend function expected type,
because it's the WHEN-case and we propagate expected type info to
the branches.
Then, after the lambdas was introduced to the containing inference
system, we're creating ResolvedLambdaAtom using the information
from analyzed lambda's shape, but didn't use known lambda resulting
type (from which we might infer FunctionTypeKind).
So, the fix is just using that already obtained information.
^KT-57446 Fixed
The only case when behavior is change is described at
computeNonTrivialTypeArgumentForScopeSubstitutor
The idea is to avoid depending on the presence of @UnsafeVariance
and instead approximate captured types in covariant argument positions
before building substitution scopes
It's correct because for Captured(*) <: Supertype,
Out<Captured(*)> <: Out<Supertype> and when we've got @UnsafeVariance
value parameters at Out, it's ok to allow passing Supertype there.
^KT-57602 Fixed
^KT-54894 Fixed
- An `OptionalAnnotationClassesProvider` only needs to be created if the
package part provider may even have optional annotation classes.
- In the IDE case, the package part provider never provides optional
annotation classes, so especially in the IDE, we can avoid a lot of
useless symbol providers with this.
In a lot of cases, when we want to process the
declaration in scope, it should be resolved to at least TYPES phase.
To avoid doing it manually in all our variety of scopes, we do it when the scope it created.
It was implicitly working manually before as lazy resolve did a lot of extra work
on resolving a declaration it was not supposed to resolve.
Now it's not the case, and we have to explicitly resolve all the declarations we need.
^KT-56543
Co-authored-by: Ilya Kirillov <ilya.kirillov@jetbrains.com>
- If the callee reference in a desugared compound assignment was an
error reference, the compound assignment wasn't cached by
`FirElementsRecorder`. This led to KTIJ-24895 (see the issue for more
details).
- The solution falls back on the compound assignment's source to decide
`isSourceForCompoundAccess`, as an operation name can be reconstructed
from the PSI.
^KTIJ-24895 fixed
We can't reduce resolve to COMPILER_REQUIRED_ANNOTATIONS phase for
annotations with arguments, because currently they don't have
argument mapping
```stracktrace
org.jetbrains.kotlin.utils.KotlinExceptionWithAttachments: By now the annotations argument mapping should have been resolved
at org.jetbrains.kotlin.analysis.api.fir.annotations.FirAnnotationUtilsKt.mapAnnotationParameters(firAnnotationUtils.kt:137)
at org.jetbrains.kotlin.analysis.api.fir.FirUtilsKt.toKtAnnotationApplication(FirUtils.kt:79)
at org.jetbrains.kotlin.analysis.api.fir.annotations.FirAnnotationUtilsKt.annotationsByClassId(firAnnotationUtils.kt:60)
at org.jetbrains.kotlin.analysis.api.fir.annotations.FirAnnotationUtilsKt.annotationsByClassId$default(firAnnotationUtils.kt:46)
```
^KT-57424 Fixed
- This optimization applied to LL FIR can be extended to K2 as a whole.
Likely, there won't be any performance benefits in CLI runs, but it's
still better to avoid useless symbol providers.
- Dependency symbol providers contained a lot of useless
`FirExtensionSyntheticFunctionInterfaceProvider`s, because such a
provider only generates function classes from compiler plugins. If
there aren't any such `FunctionTypeKind`s in the session, there is no
need to even add the provider.
- `FirExtensionSyntheticFunctionInterfaceProvider`s that are added to
the list of dependency providers despite this optimization will now be
combined in `LLFirCombinedSyntheticFunctionSymbolProvider`. It checks
`ClassId` heuristics once, which is more efficient than checking them
in every provider separately.
- Replace the "recurse one layer deep" approach of
`LLFirDependentModuleProviders` with a flattened list of dependency
symbol providers.
- This refactoring is a necessary component of supporting combined
symbol providers.
Basically, this commit reworks eraseToUpperBound* stuff
Instead of a bunch of vague flags, we introduce two modes
(rawTypes/intersection emptiness checker) which defined those flags.
Also, it makes choosing `eraseArgumentsDeeply` option always because
that how it works in K1 and also use invariant projection inside
`eraseArgumentsDeeply` for raw types for the same reason.
^KT-57198 Fixed
Jinseong Jeon