TypeAliasType seems to have the semantics of TypeAlias – can it be more like type X = Y?

[tmke8]

The following works with the new type X = Y syntax:

from typing import assert_type

class C[T]:
    type L = list[T]

    vals: L

    def f(self) -> L:
        return self.vals


def f(c: C[int]) -> None:
    assert_type(c.f(), list[int])

but it doesn't work with TypeAliasType:

from typing import Generic, TypeAliasType, TypeVar, assert_type

T = TypeVar("T")

class C(Generic[T]):
    L = TypeAliasType("L", list[T])  # error: Generic type alias within class cannot use bound type variables T

    vals: L  # error: Expected type arguments for generic type alias "L"

    def f(self) -> L:  # error: Expected type arguments for generic type alias "L"
        return self.vals


def f(c: C[int]) -> None:
    assert_type(c.f(), list[int])

I'm guessing the reason for the error is that if you do the same with TypeAlias, then at runtime, L will be generic because we specialized list with a TypeVar, but this isn't a problem with TypeAliasType, I think, so the above code should be safe?

[erictraut]

Thanks, this was an oversight in my implementation of TypeAliasType within pyright. It doesn't correctly handle the case where you specify no type parameters but reference outer-scope type parameters in the type alias definition. It should support this. It will be addressed in the next release.

[erictraut]

This is included in pyright 1.1.315, which I just published. It will also be included in this week's insiders release of pylance.