The main idea is getting rid of stub types and using just type variables
See more detailed description at docs/fir/delegated_property_inference.md
The problem with stub types is that they need really special treatment
in many places, and on the other hand, there are no clear contracts on
how they should work (that regularly leads to bugs like KT-59529)
^KT-61060 Fixed
^KT-61075 Fixed
^KT-61077 Fixed
^KT-59529 Fixed
^KT-61633 Related
^KT-61618 Related
^KT-61740 Related
^KT-59107 Related
^KT-61747 Related
^KT-61077 Related
^KT-61781 Related
Before const inlining for fields, we actually want to remember a given
field in `InlineConstTracker`. We are doing it already in const
evaluation. So we just need to ignore const inlining in
`CallAndReferenceGenerator` and delegate it to const evaluation.
#KT-60737
The original Java checker has an early return in
case of `resultingDescriptor !is JavaClassConstructorDescriptor`.
It fires if the descriptor is
`TypeAliasConstructorDescriptor`, thus further
diagnostics are not reported.
Adding these dependencies to the `api` configuration pollutes classpath for each dependant modules even if it doesn't need them. Instead, the dependencies should be declared more granularly if they're required
#KTI-1349 In Progress
The `kotlin-test` dependencies are left untouched as changing them affects publications, thus these versions are independent from the used inside our build
#KTI-1349 In Progress
There is a corresponding example inside the stdlib,
see `kotlin.text.startsWith`.
JVM and common counterpart are weakly-compatible
as the actual declaration has default arguments,
which results in `ExpectActualCompatibility.Incompatible.ActualFunctionWithDefaultParameters`
This commit allows such cases.
^KT-61732 fixed
Note that 3 tests are still muted, but for another reason: for FIR
versions of the tests, we need to compile the "pre-release library" with
the next language version which is 2.1. But since currently
LanguageVersion.LATEST_STABLE is 1.9, the compiler refuses to read
metadata of version 2.1, regardless of its own language version. Which
is correct, but it leads to irrelevant errors in the test output -- the
ones about the incompatible metadata version, NOT about the
prereleaseness.
These 3 tests can be unmuted once the default language version is
switched to 2.0.
#KT-60780 Fixed
Before the change, the compiler threw exception in the unmuted tests,
because it tried to load metadata even though it had an unsupported
version. Use the same approach as in K1 (see
DeserializedDescriptorResolver).
Now the tests are unmuted, but note that test data differs from K1. K1
does not report errors related to the class `a.A` because it loads this
class as a _Java class_, so calling its constructor and methods somehow
works. This behavior is questionable since the compiler surely knows
that it is a Kotlin class, but with an unsupported metadata version.
Trying to interpret it as a Java class may lead to subtle problems. So
it's safer for now to avoid loading Kotlin classes with unsupported
metadata versions in K2.
#KT-60795 Fixed
"Container source" means "the source of the container". So, if you
called this extension on a class (`FirClassLikeDeclaration`), it was
reasonable to assume that you would get the source of its _containing
class_, whereas the function actually returned the source of the class
itself.
Instead of inventing another name for this function, I've decided to
inline and remove it because it has only one usage.
IllegalArgumentException: class
org.jetbrains.kotlin.psi.KtLambdaArgument is not a subtype of class
org.jetbrains.kotlin.psi.KtExpression for factory
TYPECHECKER_HAS_RUN_INTO_RECURSIVE_PROBLEM
was reported when TYPECHECKER_HAS_RUN_INTO_RECURSIVE_PROBLEM was
reported on a FirLambdaArgumentExpression, report it on its
`expression` instead.
#KT-60380 Fixed
There was only one real inheritor of AbstractFirBaseDiagnosticTest,
so there is no need in the whole hierarchy of abstract test runners
Also AbstractFirOldFrontendLightClassesTest is not intended to check any
diagnostics or CFG dumps, so this logic was removed (they still can be
checked using non-legacy AbstractFirDiagnosticTest from the main test
infrastructure)
^KT-61640 Fixed
This change doesn't affect any tests in the moment, because it's part of
bigger refactoring of proper storing IR declarations during FIR2IR
conversion (KT-61637)
Without this change, there is one test break in branch for KT-61637:
- FirPsiBlackBoxCodegenTestGenerated.Multiplatform.K2.testJavaMethodWithTypeParameter
This is needed for two reasons:
1. common and platform modules are analyzed in the same setup with same
dependencies and configurations, so the results of the enhancement for
any function will be completely identical. So by sharing enhanced
symbol storage, we avoid recomputation enhanced functions and improve
performance
2. There is a goal to have some unique key for IR declarations in FIR2IR.
For regular declarations, the key is just Fir symbol, and for fake-overrides
it is a pair of original symbol and fake override owners lookup tag.
But for enhanced functions the symbol is not unique, because we create
different symbols for the same enhanced function for different modules.
So this change fixes this problem
^KT-60397 Fixed
This change doesn't affect any tests in the moment, because it's part of
bigger refactoring of proper storing IR declarations during FIR2IR
conversion (KT-61637)
Without this change, there are some test breaks in branch for KT-61637:
- FirPsiBlackBoxCodegenTestGenerated.Multiplatform.testStarImportOfExpectEnumWithActualTypeAlias
- FirPsiBlackBoxCodegenTestGenerated.Multiplatform.K2.DefaultArguments.testNestedEnumEntryValue
- FirPsiBlackBoxCodegenTestGenerated.Multiplatform.K2.Basic.testEnumEntryNameCall
UncaughtExceptionPath edges are used to influence smart-casting within
catch and finally blocks. Previously these edges were added from every
node which could throw an exception. But only assignment nodes influence
smart-casts by resetting inference back to some less specific type.
Therefore, instead of tracking every possible node which could throw an
exception - even though almost every statement node can - only add edges
from assignment nodes to catch and finally blocks. This fixes many
missing exception cases and also reduces the total number of incoming
edges to catch and finally blocks.
#KT-56872 Fixed
Optimization in 2439c22ff6 introduced an array `absentArguments` with
default values of all parameter types, which is copied in `callBy`
instead of being recomputed each time. Unfortunately, its size was
computed incorrectly: `maskSize` should only take into account
value parameters (see the `parameter.kind == KParameter.Kind.VALUE`
check in `callBy`).
This led to an issue where if the function had 32 * N value parameters
and 1 receiver parameter, `maskSize` was greater by 1 than expected,
which caused an exception due to mismatching argument array sizes.
#KT-61304 Fixed
There is a case when we shouldn't hide a function even if it overrides
builtin member with value parameter erasure: if substituted kotlin
overridden has the same parameters as current java override. Such
situation may happen only in case when `Any`/`Object` is used as
parameterization of supertype:
// java
class MySuperMap extends java.util.Map<Object, Object> {
@Override
public boolean containsKey(Object key) {...}
@Override
public boolean containsValue(Object key) {...}
}
In this case, the signature of override, made based on the correct
kotlin signature, will be the same (because of { K -> Any, V -> Any }
substitution for both functions). And since the list of all such
functions is well-known, the only case when this may happen is when
value parameter types of kotlin overridden are `Any`
This change doesn't affect any tests in the moment, because it's part of
bigger refactoring of proper storing IR declarations during FIR2IR
conversion (KT-61637)
Without this change, there is one test break in branch for KT-61637:
`FirPsiBlackBoxCodegenTestGenerated.SpecialBuiltins.testMapGetOrDefault`
^KT-61573 Fixed
Review: https://jetbrains.team/p/kt/reviews/11969/timeline
Tests:
- MultiPlatformIntegrationTestGenerated
- CliTestGenerated
- MultiPlatformIntegrationTestGenerated
- DiagnosticTestGenerated.Multiplatform
- FirLightTreeOldFrontendDiagnosticsTestGenerated
Also add -Xexpect-actual-classes flag to all necessary ./libraries/* modules
Otherwise compilation of those modules failes because of `-Werror`
`knownVariables` can't be moved to the outer
function, because it's specific to the current
path: we want to check that the contract
condition is approved along the current path.
For example, in the
`FirPsiDiagnosticTestGenerated.testPropertyGetter`
test when analyzing the function, we want
to check if we can derive the `d1 == Any`
type statement (where `d1` is the receiver).
If we can't, the flow won't contain `d1`
among `knownVariables`
- `getConstructorKeyword` forces an AST load due to a child search. In
most cases, we don't need to check the constructor keyword in primary
constructors at all (e.g. `class ABC()` with an implicit primary
constructor). The keyword check is only needed when there are primary
constructor modifiers (e.g. `class ABC constructor()`).
Thanks to Dima Gridin for the suggested fix.
^KT-61635 fixed
Denis.Zharkov