- Change back 'TestVersions.AGP' to be 'object class' with constants
inside, so they could be referenced inside version annotation
- Fix issue when AGP minimal version could be higher than AGP max
version
^KT-45747 Fixed
This is a partial revert of commit bc1c2ada
We have to use old type intersector for emptiness check because
KT-53656 were found. We return to old type intersector usage also
for type parameter bounds to prevent other possible problems.
Since Kotlin 1.7.0 plugin is tracking each java file individually in
task, rather than java SourceSet root directories. Such approach removes
requirement to pass them via `-Xjava-source-roots` compiler argument.
^KT-53390 Fixed
After an accidental change in 846537b367, we started generating metadata
into anonymous classes for callable references. This metadata contained
the Kotlin representation of the referenced function. In KT-53794, this
led to a problem where Java's protected visibility could not be
represented in Kotlin terms, which crashed the backend.
But also, this metadata is useless because there's no real use case for
interpreting it (since you already have a function reference object at
runtime with all the needed information), so it would take some extra
space in the bytecode.
#KT-53794 Fixed
The default type checker doesn't have a correct type refinement setup.
This can cause false positives in subtyping of upper bounds in edge
cases with expect type arguments.
See the tests in the intellij repo.
KTIJ-22295
This intrinsic was always calling the supplied suspend function
non-virtually, and this was hardcoded in the intrinsic generator.
Now this can be configured with `superQualifierSymbol` in the respective
IrCall, just the way this works with any other calls.
As a side effect, global optimizations now process this intrinsic more
correctly (before that, they didn't take the intrinsic generator into
account, and treated the call as virtual).
Previous implementation of startCoroutineUninterceptedOrReturn expected
that the receiver (i.e. the suspend function object to start) of type
SuspendFunctionN (i.e. suspend function type with arity = N) is also an
instance of FunctionN+1 (i.e. regular function type with arity = N + 1)
with proper convention.
While compiler tries to achieve that (for better compatibility with
Kotlin/JVM, by generating additional supertypes and methods for classes
implementing suspend function types), this doesn't work e.g. for
implementations defined in Swift.
Fix this by adding fallbacks for cases when the receiver is not
a FunctionN+1, while keeping the fastpath for default FunctionN+1
cases.
^KT-51043 Fixed
Anton Lakotka