False positive with splat and explicit kwarg

[torfsen]

Bug Report

Mypy reports multiple values for a kwarg if args are given via splat and the kwarg is explicitly set.

To Reproduce

def foo(x: int, y: int, z: str) -> None:
    print(x, y, z)

args = [1, 2]
foo(*args, z="...")

Expected Behavior

Mypy should not report any problems.

Actual Behavior

mypysandbox.py:5: error: "foo" gets multiple values for keyword argument "z"  [misc]
mypysandbox.py:5: error: Argument 1 to "foo" has incompatible type "*list[int]"; expected "str"  [arg-type]
Found 2 errors in 1 file (checked 1 source file)

Your Environment

$ mypy --version
mypy 1.19.1 (compiled: yes)

$ python --version
Python 3.12.12

Mypy config from pyproject.toml:

[tool.mypy]
strict = true
allow_redefinition =  true

[torfsen]

This is very similar to #10481, but here it's splat + kwarg instead of splat + splat. I wasn't sure if this difference is relevant, feel free to close as duplicate.

There is also #2582, but that is very old and has long been fixed.

#10348 looks like it's the same issue as this one and #10481 but reported just for datetime.datetime.

And then there is #6799, which is also very similar.

So definitely some duplicates, but I don't know enough about the details to figure out which one should be the main ticket.

[A5rocks]

FWIW I'm tempted to say this isn't a bug, because the list could technically be 3 items long and mypy wouldn't know any better. Let's see what others think...

[randolf-scholz]

In the provided example, using a tuple instead of a list fixes the issue. The question is what to do in a case like

def foo(x: int, y: int, z: str) -> None: ...

def test(items: list[int]) -> None:
    foo(*items, z="string")

FWIW ty and pyrefly also error here, while pyright doesn't There is a trade-off:

1. First match the keyword argument, and match the positionals afterwards. This avoids the false positive if the list happens to be the correct length, but produces a false negative if the list has incorrect length
2. First match the positional, and then the keyword args (this is mypy currently does), which produces a false negative when the list happens to have the correct length.

[sterliakov]

I'm +1 on keeping this as-is - passing arguments like this is a relatively common source of bugs, just as bad as func(**kwargs, foo=bar). Fixed-size tuples should be allowed (they provide enough information), but not an arbitrary iterable.

But current mypy behavior is inconsistent - we do not report too few arguments for func(*args), for example. We need to agree where the line is drawn and align all such too many/too few args checks for *args and **kwargs behavior.