Incompatibilty Error from Base Classes with **kwargs

[adam-grant-hendry]

Related to this SO question, I'm trying to understand why

• this doesn't yield errors

but

• adding **kwargs to the end of the run functions does

[adam-grant-hendry]

I believe, if I'm not mistaken, the error could be due to the potential for a keyword collision, which I gleaned from PEP 692. Confirmation from the community would be appreciated though!

e.g. If I made the name of the 4th arguments the same, each could fail the same way if I tried to pass foo as a keyword argument:

>>> a = Foo()
>>> a.run(1, 2, 'baz', foo='bar')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: func() got multiple values for argument 'foo'

but if I keep the argument names different, then the above would work if I tried to call Bar.run(1, 2, 'baz', foo='bar'), but break for Foo.

[JelleZijlstra]

I don't think there's a very principled answer here: strictly speaking both should be an error. I believe pyright will flag both as errors but I haven't tried.

Mypy is fairly lenient with the names of positional parameters in overrides, in part because mypy's design predates convenient syntax for positional-only parameters. That's why your first example doesn't give an error.

[adam-grant-hendry]

@JelleZijlstra Thank you for your reply.

I can copy-paste the examples to the pyright playground and see what they say.

This is a source of confusion for me and others (see this SO post). If you would be able to weigh in, it would be much appreciated!

[adam-grant-hendry]

@JelleZijlstra On the pyright playground, the OP code has no errors, which is what I would expect since nothing is statically typed in that code so everything should be read as type Any.

Is this by chance a mypy bug? Should I file an issue from this discussion?