BUG: Fix agg with single-element function list given arg and kwargs

[luke396]

• closes BUG: agg ignores args/kwargs when giving a list-like argument #50624
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/v1.5.3.rst file if fixing a bug or adding a new feature.

[rhshadrach]

Thanks for the PR! Unless it's a regression or some exceptional circumstances, we target the next minor version for the whatsnew. Right now, that's 2.0.0.

[rhshadrach]

This is going to fracture the behavior. Currently all list-like func are treated the same - it is the wrong behavior, but it's at least consistent. I think we need to find a better solution for passing through args and kwargs.

We save the args and kwargs to an instance on L112-113. Can we use them when we do the compute down below?

[luke396]

Yes, we do need a better solution. I am going to try to solve this with their instance on L112-L113.

There is also the question of whether we should now focus on solving a single-elements function list, or should we consider the case of more elements.

For more elements, my initial approach was to use something like inspect.getargspec or inspect.signature. But if the parameters between several functions have the same naming or other conflicts, the problem seems to be more difficult.

I haven't fully sorted out my thoughts yet, so I've come up with a somewhat strange solution to a single-element list.

[rhshadrach]

I don't think we should only fix the single-element case as this would further fragment behavior for users. We should either fix all cases or leave behavior as-is.

For more elements, my initial approach was to use something like inspect.getargspec or inspect.signature. But if the parameters between several functions have the same naming or other conflicts, the problem seems to be more difficult.

I believe you're thinking of only passing arguments if the function accepts them. I don't think we should do that - consider in particular the case where a function accepts *args or **kwargs. Also, a user might have

def func(x, myarg=1):
    ...

and call ....agg(func, my_arg=2). This code should raise - the user has a typo. Instead it would silently produce incorrect results.

[luke396]

I see, I did not consider the typo problem, thanks for pointing it out!

I will try to solve the multi-element list directly, but it may take a little time.