Incorrect inference with ambiguous overloads with mutiple arguments

[randolf-scholz]

Summary

https://play.ty.dev/7e986f4b-4688-4a6f-951b-f78ef9eebb09

from typing import Any, overload, assert_type

class A[T]:  # covariant
    def get(self) -> T: ...

@overload
def op(l: A[None], r: A[None]) -> A[None]: ...
@overload
def op(l: A[None], r: A[Any]) -> A[None]: ...
@overload
def op(l: A[Any], r: A[None]) -> A[None]: ...
@overload
def op(l: A[Any], r: A[Any]) -> A[Any]: ...

def test(x: A[None], y: A[Any]) -> None:
    assert_type(op(x, x), A[None])  # spec: ✅️ 
    assert_type(op(x, y), A[None])  # spec: ✅️
    assert_type(op(y, x), A[None])  # spec: ✅️
    assert_type(op(y, y), Any)     # spec: ❌️ (expected Any)

ty incorrectly infers A[None] for op(A[Any], A[Any]). See python/typing#2196 and #665 for context.

Single argument ambiguous overloads seem to currently infer as Unknown.

Version

No response

[carljm]

Thanks for the report! It looks like pyright and pyrefly share the same behavior as ty here, while mypy/zuban infer A[Any] for the last three assert_type. So it seems no type checker is quite in compliance with the spec here currently.