reportMatchNotExhaustive false positive for tuples

[InSyncWithFoo]

Minimal reproducible example (playground):

def f(l: tuple[int, ...]) -> int:
    match l:  # Cases within match statement do not exhaustively handle all values. Unhandled type: "tuple[int, ...]"
        case []:
            return 0
        case [head, *_rest]:
            return head

This is almost identical to #6843, but the reason for its closure doesn't apply here. Is this a bug?

[erictraut]

This is expected behavior, not a bug. Match exhaustion is detected via type narrowing. In your example, the first case statement (with pattern []) does not narrow the subject type in the fall-through case. Its type remains tuple[int, ...]. For comparison, mypy's behavior is the same.

[qexat]

In the case where the first pattern is (), couldn't it narrow to tuple[int, *tuple[int, ...]]?

[qexat]

In the case where the first pattern is (), couldn't it narrow to tuple[int, *tuple[int, ...]]?

After some thinking, I'm not sure that it would play well with more complex types, and there are probably some non-obvious edge cases that would prevent this narrowing that I haven't thought about