When BI/incomplete call is present in return argument, it will be
added to the main call-tree, leading to requirement violation in
`ConstraintSystemCompleter.getOrderedAllTypeVariables`
as after `FirBuilderInferenceSession.inferPostponedVariables` it will
be completed, and its type variables couldn't be found anymore
In K1, we actually do the same, but we are able to find type variables
of such calls due to architecture difference:
In K2, `getOrderedAllTypeVariables` uses FIR as the main source to
lookup
In K1, `getOrderedAllTypeVariables` uses resolution atom tree, where
candidate/CS of the call is still available after
`inferPostponedVariables`
In fact, all incomplete calls were analyzed in FULL mode according to
the contract of `FirBuilderInferenceSession.addPartiallyResolvedCall`.
Thus, it means we normally shouldn't add them to the main call-tree, but
accidentally do it as incomplete calls contain non-completed candidate
This particular commit addresses the problem partially, only
in cases when the expected return type for the lambda is Unit and when
the incomplete call is located in the last expression of the lambda
In such cases, we can skip the call from the last expression completely,
since all potential calls there were analyzed in FULL mode,
and couldn't introduce any useful info to the CS of the main call-tree
^KT-54294
This commit adds missing pieces for the puzzle:
Annotation instantiation feature uses IrProperty's initializer to instantiate
properties from other modules that have default values which weren't
specified on call site.
To support this feature properly, Fir2IrVisitor should fill LazyIrProperty's
backing field initializer with information from Fir.
To get this information into Fir, FirMemberDeserializer should be able to read
it from KotlinJvmBinaryClass with AnnotationLoaderVisitorImpl. (klibs are unsupported for now)
There's a catch with enum entries references: we can't access session.SymbolProvider to resolve it
because we're still at the deserialization stage, and it can cause StackOverflow if enum is nested in the
same class (see RequiresOptIn.Level). To mitigate this, a new FirEnumEntryDeserializedAccessExpression is produced
instead; it is later replaced with the correct reference in the Fir2IrVisitor.
^KT-58137 Fixed
Also add test to loadJava folder with annotations default values that
verifies metadata loading
Java declarations do not have a file parent, their outer element is
IrExternalPackageFragment.
The test did not fail in K1 only by accident, because there's an extra
IrTypeOperatorCall on the property reference's dispatch receiver, which
automatically makes it non-inlinable. In K2, it's IrGetField, so we were
trying to compute if its dispatch receiver is inlinable and for that we
tried to get its fileParent, which resulted in exception if it comes
from Java.
Also refactor PropertyReferenceDelegationTransformer.canInline a bit.
#KT-57955 Fixed
Previously, when a candidate was found with an applicability that is
better than the current best applicability, all previous candidates were
thrown away. Now we keep them, unless the new applicability is
successful. If no successful candidates are found, we fully resolve all
the unsuccessful ones and select the ones with the least bad
applicability. This improves diagnostics for unresolved calls.
#KT-57844 Fixed
The code in IrTypeMapper was incorrectly translated from
KotlinTypeMapper during the development of JVM IR. The
`classDescriptor.hasBigArity` condition in KotlinTypeMapper was checking
if the class represents a function or a suspend function with big arity,
and the suspend function part was lost during conversion.
This resulted in incorrect generic signature being generated, which led
to malformed type exceptions from reflection, and compilation errors
from kapt stub generation.
Also, change a comment in irCodegenUtils to avoid confusion of numbered
function types (kotlin.jvm.functions.Function1, ...) with the big-arity
type kotlin.jvm.functions.FunctionN.
#KT-58375 Fixed
Basically, the test checks that adding Enum.entries feature doesn't
break the existing code where it clashes with the user-defined "entries"
declaration; it's better to have a black-box test to be sure
that the compiler doesn't invoke something different at runtime
The test covers KT-53153 and KT-56587
Merge-request: KT-MR-9798
Merged-by: Michail Zarečenskij <Mikhail.Zarechenskiy@jetbrains.com>
Previously, updateTypeInBuilderInference was calling
updateTypeFromSmartcast to actually perform type update in
implicit receiver after stub types was inferred
Such action results in creation of following FIR:
FirSmartCastExpression(
original=FQAE(FirImplicitThisReference, typeRef=R|Inv<Stub>|)
typeRef=R|Inv<String>|
)
in receiver position during completion of calls
However, it wasn't the case in general situation due to
action of FirStubTypeTransformer, which, in turn visits and updates
type ref inside original expression, but only if there was at
least one call (that was completed) using that implicit receiver
As after such type update updateTypeFromSmartcast function does
nothing
Yet in situation, when there was only partially resolved calls
referencing that implicit receiver we actually create smart-cast
expression and don't update type
The change just removes usage of updateTypeFromSmartcast and
replaces is with direct type update
We still mutate state of implicit receiver, potentially
improperly, it should be addressed in future
^KT-54708
^KT-58365 Fixed
These two functions apparently are represented in Kotlin as methods
of `kotlin.String`. Because of that we accidentally treated them as
builtins.
To also minimize such cases, added filtration by return type. We are
allowing to interpret only these functions that have primitive or
unsigned return type.
#KT-57028 Fixed
Do not check that non-indy-generated lambdas inherit from Lambda,
because when we enable indy lambdas by default, we're also changing the
behavior of non-indy lambdas so that they won't have the Lambda
superclass anymore (see KT-45375). Since this test was added to check
that any annotation disables indy lambda generation, just leave those
lambdas and rely on the CHECK_BYTECODE_TEXT directive to verify that
there are no LambdaMetafactory in the bytecode text.
Most of these tests check the specific structure of lambdas when they
are generated as classes, and they start to fail once invokedynamic
lambdas are enabled by default.
Use -Xlambdas=class in tests which were checking specific things related
to how anonymous classes for lambdas work (such as receiver mangling,
function arity etc.)
This is needed to correctly handle the case, when we have the same java
class in common and platform module. In this scenario we have two
different classes on frontend (because symbol providers are not shared)
and completely different enhanced function in scopes of those classes,
but exactly one IrClass for those classes, which is cached during
conversion of common module together with cache of its fake overrides.
So using FirDeclaration as a key to FO cache leads to the problem, when
we can not find cached value for platform module, because it has different
fir declarations for the same real decalration
^KT-58030 Fixed
Before this commit, for property candidates in K2 their types wasn't
inferred/susbtituted properly.
So, when candidate for fooBar.liveLoaded.invoke() was created,
the type of `fooBar.liveLoaded` was just X type parameter for which
there is no any `bar()` functions in its member scope.
While proposed semantics is a bit different from K1, where
both property and invoke candidates are united into common system,
it doesn't contradict to the specification (https://kotlinlang.org/spec/overload-resolution.html#callables-and-invoke-convention)
which says explicitly that invoke-convention should be desugared as
`r.foo.invoke()`, thus `r.foo` should be completed independently.
Also, this strategy supports some reasonable use-cases like KT-58259
while it's still a breaking change but for more artificial-looking
situations (see KT-58260) and should be passed through
the language committee.
The changes in stubTypeReceiverRestriction* tests looks consistent
because of how `genericLambda` now works
(with full completion of property call).
NB: The code is going to be red once KT-54667 is fixed and also there's
already similar diagnostic in K1 (INFERRED_INTO_DECLARED_UPPER_BOUNDS)
^KT-58142 Fixed
^KT-58259 Fixed
^KT-58260 Related
Alexander Udalov