Previously, there was receiver of type Nothing, which could case result
of the call to be inferred to Nothing too. This could case bogus
UNREACHABLE_CODE diagnostics in cases like this:
```
fun <T> id(x: T) = x
fun test() {
id(unresolvedReference) // type of statement is 'Nothing'
// ... everything here is marked as unreachable ...
}
```
This commit changes type of receiver for such calls form Nothing to
ErrorType.
This commits introduces testdata changes, where NI behaviour strictly
improved, after several previous fixes.
For some tests, just WITH_NEW_INFERENCE directive was added. It
indicates, that some of previous commits first introduced error in that
test, and then some other commit fixed it (netting no overall testdata
change). It is preferrably to keep those annotations until we will
migrate to NI completely, to prevent unexpected regressions.
This commit support the following case.
Suppose we have such declaration:
fun <T> foo(): T { ... }
Then in code we want to use it like this: `foo() as String`.
But in LV <= 1.1 we have type inference error: "Not enough
information for type parameter `T`". This error happened because we
do not use type from cast as expected type for call.
In this commit we fix this problem and use this type as expected type
in following cases:
- our function has only one type parameter (this can be relaxed later)
- function parameter types and extension receiver type not contains `T`
Also this fix problem with `findViewById`.
Already signature was: `fun findViewById(...): View`
and was used like: `findViewById() as MyView`.
New signature is `fun <T : View> findViewById(...): T`
and old usage was broken because of problem described above
We prefer denotable types when we solve constraint system.
I.e. if for T we have not equality constraint with captured type we can approximate captured type to denotable type.
There is no need to report an error in case of non-top-level
capture conversion, just don't add relevant capture type in the system instead
If system can be solved successfully without captured type, then it's just fine
(see KT-13950)
In case of contradiction TYPE_INFERENCE_PARAMETER_CONSTRAINT_ERROR is reported
#KT-13950 Fixed
It helps to catch TYPECHECKER_HAS_RUN_INTO_RECURSIVE_PROBLEM errors
because they appear after return type calculation began
#KT-6271 Fixed
#KT-3272 Obsolete
Instead of erasing descriptors with conflicting substitution,
use invariant CapturedType(<projection>) as replacement for type parameter
within default member scope.
After substitution leave such types 'as is' everywhere except return types,
use common approximation for them.
#KT-9294 In Progress
#KT-5411 Fixed
#KT-8647 Fixed
#KT-9462 Fixed
#KT-9893 Fixed
#KT-7581 Fixed
#KT-7296 In Progress
Basically it's wrong to use original types' member scope
as a worker for SubstitutionScope.
Member scope should always be determined by type constructor's default one
and substitution/arguments
#KT-10448 Fixed
Kirill Rakhman