R0913 excludes keyword- and positional-only arguments

[charliermarsh]

Bug description

Pylint's too-many-arguments rule seems to exclude keyword- and positional-only arguments from the "total argument" count:

def f(a, b, c, d, e, *, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t):
    ...


def g(a, b, c, d, e, f):
    ...

Configuration

No response

Command used

pylint a.py

Pylint output

************* Module a
a.py:5:0: R0913: Too many arguments (6/5) (too-many-arguments)

Expected behavior

My instinct is that it should include keyword- and positional-only arguments, but I'm wondering if this is intentional.

Pylint version

pylint 2.17.2
astroid 2.15.3
Python 3.11.1 (main, Dec 23 2022, 09:25:23) [Clang 14.0.0 (clang-1400.0.29.202)]

OS / Environment

No response

Additional dependencies

No response

[mbyrnepr2]

Thanks for this @charliermarsh! I believe that the too-many-arguments checker predates positional-only arguments and explains why they were not originally handled by the checker. I don't think there is any reason to exclude the positional-only and keyword-only parameters from the check.

[parched]

@charliermarsh @mbyrnepr2 This going to throw up a lot of warnings in my code. I believe the intention of this rule is to ensure args don't get mixed up. That can't happen with kwargs, as they're explicit. I do believe position-only args should be included though.

[Pierre-Sassoulas]

I think that too-many-arguments is a refactoring check and its intention is only to limit the number of args of a function (i.e. you should create a data structure to group semantically close arguments together). We asked ourselves if we should increase the default value from 5 to 7 though, there's indeed a lot of new message raised. Not having too many positional argument because you don't know what is what is another check not yet implemented (#385)