There are two different forms of types intestion:
1. Type parameters with multiple bounds
2. Smart casts
The problem was that when member scope of type intersection contained
effective duplicates and that lead to overload resolution ambiguity in
strange cases like `x.hashCode()`
For first type we do effectively the same thing as when building member
scope for class extending several interfaces: group all descriptors by
both-way-overridability relation and then choose most-specific in each
group.
For smart casts we do basically the same thing but with special
treatments:
1. From all descriptors that _equal_ to most specific we choose
the one that works without smartcast if possible (i.e. we choose first from candidates list)
2. If smart-cast value seems to be unstable we use only member scope
of receiver type + all descriptors from smart cast possible types
that has incompatible signature. If we'd include all of them and
choose one as more specific, and it would lead to false
SMART_CAST_IMPOSIBLE (see test unstableSmartCast.kt)
#KT-3996 Fixed
#KT-10315 Fixed
In cases when signature of special bridge is the same as current method,
but type is not 'Any?'.
Also there is tiny optimization:
only null check needed if value parameter type is mapped to Object,
but it's not nullable.
#KT-9973 Fixed
- base class method wins against a (default) interface method,
so an abstract base class method should always be implemented
in a derived class;
- interface methods clash regardless of abstract/default
with possibly undefined behavior at run-time,
so a class or interface should always define its own method
for methods inherited from multiple interfaces and not from base class;
- meaningful diagnostics for class inheriting conflicting JVM signatures.
Since no override will happen under Java 8 rules,
ACCIDENTAL_OVERRIDE is misleading for this case;
- update testData.
This proved to be a fragile technique, which probably doesn't even improve
performance in most cases but has lots of unexpected problems: unconditional
initialization of reflection classes, increasing the size of the bytecode, bugs
with <clinit> in annotations on JVM 6, inability to support conversion of a
class from Kotlin to Java without recompiling clients which use it
reflectively, etc.
0. Such properties are called special because their accessor JVM name differs from usual one
1. When making call to such property, always choose special name
2. When generating Kotlin class inheriting such property generate `final bridge int size() { return this.getSize(); }`
3. If there is no `size` declaration in current class generate `bridge int getSize() { // super-call }`
This format of the string table allows to reduce the size of the Kotlin
metadata in JVM class files by reusing constants already present in the
constant pool. Currently the string table produced by JvmStringTable is not
fully optimized in serialization (in particular, the 'substring' operation
which will be used to extract type names out of generic signature, is not used
at all), but the format and its complete support in the deserialization
(JvmNameResolver) allows future improvement without changing the binary version
