Instead of loading the descriptor and checking equals(), we now check if the
annotation is in fact "suppress" by simply comparing its FQ name. This allows
us to suppress warnings while compiling built-ins: the problem is that
built-ins have its own "suppress" annotation class (defined in the sources)
which differs from the one loaded by the compiler (from kotlin-compiler.jar)
Move from package "kotlin.reflect" to "kotlin.reflect.jvm.internal". They are
internal detail of the compiler and should not be used directly (especially now
that "kotlin.reflect" is in default import paths).
Also rename "KFunctionImplN" to "KFunctionNImpl", because this name makes more
sense
SAM-related code in codegen was using JavaClassDescriptor directly, which has
an erased signature. Create and use a new abstraction SamType which has a full
generic signature of a type which was used in the SAM construct
Metadata for KFunction classes is now longer serialized along with built-in
classes. This effectively means that it's no longer possible to find KFunction
classes via dependency on built-ins. There should be a kotlin-runtime library
in the specified classpath for reflection types to be resolvable.
A lot of tests were moved and changed, because tests on callable references
require stdlib in classpath from now on
- delete fields which were supposed to be used for caching, but mistakenly were
never used
- delete/inline unused methods and constants
- delete useless logic related to root package fragment: package fragment
providers are only supposed to find packages they're aware of
Create all property metadatas in <clinit> and save them to a static array
$propertyMetadata. Getter/setter of each delegated property will then just
obtain the corresponding instance from that array
#KT-4232 Fixed
OverridingUtil will be present at runtime and we should strive to minimize the
size of our runtime classes
Also add a comment to getOverriddenDescriptors()
Fix warnings, simplify code. Untangle a very complex code related to
diagnostics on multiple default values and different names of parameters in
supertypes; fix a bug there, add test
- the member and the container were mixed up
- don't output "defined in" part of the message, because we tell later where
it's already defined in
- fix an instability of the diagnostic order in OverloadResolver by using
LinkedHashSet
Svetlana Isakova