Basically, this commit reworks eraseToUpperBound* stuff
Instead of a bunch of vague flags, we introduce two modes
(rawTypes/intersection emptiness checker) which defined those flags.
Also, it makes choosing `eraseArgumentsDeeply` option always because
that how it works in K1 and also use invariant projection inside
`eraseArgumentsDeeply` for raw types for the same reason.
^KT-57198 Fixed
Before this commit, we added Enum.entries only in case when
LanguageFeature.EnumEntries was ON (with an exception in K1/Java case).
In this commit we add Enum.entries unconditionally, and in case
the language feature is OFF we filter them out during tower resolve.
If some java class has multiple supertypes then we need to collect
overriddens from all those types directly, even if superTypeScope
(which is FirTypeIntersectionScope in this case) returns only
one symbol from one of this types (not intersection one)
This is needed to proper enhancement in cases when some type occurs
multiple times in supertypes graph with different nullability
of arguments:
class ConcurrentHashMap<K, V> : AbstractMap<K!, V!>, MutableMap<K, V>
If we try to find method `get(key: K): V` supertype scope returns
`AbstractMap.get(key: K!): V!` (because it actually overrides
`MutableMap(key: K): V?`), but we need to get both symbols to
properly enhance types for `ConcurrentHashMap.remove`
1. Inner class constructor should have its outer class as a dispatch
receiver, since it is necessary for the call. Before it was null
2. Substituted inner class constructor should have its original dispatch
receiver type with the proper substitution. Before it was set to the
class itself (since the class was usually passed as a new dispatch
receiver)
Also, modify FIR renderer, so it properly renders the dispatch receiver
of the constructors
Because those test are single Java files, many of them use a top-level
class as a container for multiple classes to be tested. Such tests do
almost nothing if those nested classes aren't handled.
- Add utilities to add new attribute to ConeAttributes
- Get rid of FlexibleNullability attribute (it can be easily inferred
for any flexible type at any moment)
- Fix determining of EnhancedNullability attribute
Some of the changed data is correct, but some diagnostics are incorrect
Corner cases like having both contains(Object) and contains(String)
within implementation of Collection<String> is not supported
Kirill Rakhman