Repository navigation
Regression: false positive about overload item never being matched #16338
Description
This fragment generates a false positive, starting from #15913:
from typing import overload, Any
class C:
@overload
def f(self, e: str, **kwargs) -> str: ...
@overload
# error: Overloaded function signature 2 will never be matched: signature 1's
# parameter type(s) are the same or broader
def f(self, *args, **kwargs) -> Any: ...
def f(self, *args, **kwargs):
passThe signature of item 2 is wider than item 1 due to accepting an arbitrary number of positional arguments and with arbitrary types, so this is a false positive.
cc @ilevkivskyi
Activity
ilevkivskyi commentedon Oct 27, 2023
This is one of those case when I ask myself how did this ever work, and it turns out it didn't. It turns out having
(*args: Any, **kwargs: Any)as a fallback overload caused false positives almost always. So this is not really a regression, now mypy is consistently broken, LOL.Anyway, the fix is in #16343, it turns out we never propagated
is_proper_subtypeflag tois_callable_compatible(). This is a major change, let's see how bad the fallout will be.Properly use proper subtyping for callables (#16343)