The reference type is approximated in Psi2Ir, so we may get Nothing as
a reference type argument. Better look at the arguments of the
referenced function.
The SAM adapter is generate on declaration site. This is different
from the JVM approach.
`external fun interface` is banned for now.
Reusing interface declaration for the adapter is a hack which
reduces code size and makes importing/exporting the adapter
effortless.
Otherwise:
* should the dispatch receiver of a call be another call to a `suspend
fun` wrapped in something that is optimized away later, the owner of
the method will be incorrect;
* references to functions returning non-Unit but casted to `() ->
Unit` (allowed by new inference) might in fact not return Unit after
tail call optimization.
`invoke` in suspend lambdas overrides FunctionN.invoke, so the
refactored BridgeLowering already generates correct bridges there.
All the hack does is break overrides of interface suspend methods.
values within initializer blocks.
The issue occurs in code like this:
```
class C {
var b = true
init {
b = false // Missing PUTFIELD for this statement
}
}
```
Added a new statement origin for field initialization (at declaration)
instead of relying on `origin == null` in ExpressionCodegen to determine
whether to generate the initializations.
This was unintentionally broken in
d68a1898d0.
- [JS IR] Unmute fixed tests
- [IrText] Update testdata
- [WASM] Temporary turn wasm test off
- [FirText] Temporary turn fit text tests off
- [JVM IR] Turn off klib jvm test
- [IR] Add new test
Special bridges which are generated as a replacement for a non-final
fake override and use invokespecial to call the already existing
implementation in the superclass, should not be final. Otherwise we
can't generate an override for the original Kotlin function in a
subclass with the same JVM signature.
backends by removing IMPLICIT_NOTNULL casts from IrStringConcatenation
arguments.
Also fixed an issue where IrStringConcatenation can be lowered into
a null String instead of a literal "null" String if the lone argument
was a platform type String or String with enhanced nullability and the
value was null (e.g., "${FromJava.nullPlatformString()}").
The fact that this was a codegen box test was probably an abuse of the
testing infrastructure. It was never meant to pass, since it checks
presence of an error, and diagnostic tests aren't enough since they
might not check the full compiler execution.
This leads to problems after 4dd794c2d2, because the immediate super
function's DefaultImpls and the implementation's DefaultImpls have
differing type parameters.
Looks like resolveFakeOverride was used here (maybe unintentionally) as
a workaround to the problem caused by the incorrect origin check in
isDefinitelyNotDefaultImplsMethod.
Otherwise EnhancedNullability annotation instance is lost (for some
reason it's not translated to IR elements), and in the newly added test,
the wrapper type in Java is confused with primitive, and this causes
a platform declaration clash error.
Also make IrTypeArgument.eraseTypeParameters private, since it's not
used outside.
This fixes an issue with lateinit properties where the metadata from
the original field was not copied to the nullable field in
LateinitLowering. Also consolidated related tests.
Steven Schäfer