False negative - assign `list.reverse()` instead of `reversed(list)`

[orSolocate]

Current problem

Parallel issue (being 'brought back from the grave') - #5722.

when you use the reverse() method on a list object, it modifies (reverses) the list and returns None.
A common mistake is to assign the return value of this assignment (i.e. None) to a variable thinking you assigned the sorted list instead.

lst=[3,2,1]
A=lst.reverse()

Results with:
None in A and [1,2,3] in lst.

The problem is that Pylint does not detect that, running this snippet shows no messages.

Desired solution

When this case is detected, a message should be emitted explaining the user to use the built-in reversed() function instead.
Issue is a parallel case of issue #5722, using the assignment-from-none message looks proper to me.

Additional context

There might be a compatibility issue with detecting the list.sort() and list.reverse() methods in newer version of python, where pylint sees the stub function (that contains only pass) and thinks it is a function with no return, emitting the assignment-from-no-return error message. When running pylint on the following snippet on Python3.8 and Python3.10 I got:

Snippet:

lst = [1, 2, 3]
rev_lst = lst.reverse()
sort_lst_err = lst.sort()

Python3.8: got assignment-from-none for the "lst.sort()" line.
Python3.10: got assignment-from-no-return for both lst.reverse() and lst.sort() lines.

Is it a false-positive? or did I do something wrong?

[orSolocate]

I could work on it if we decide we want to have it.

[Pierre-Sassoulas]

Thank you for opening the issue. I think the first thing to do for assignment-from-none is to follow Jacob's suggestion here: #7853 (comment) : it's going to have a big impact maybe even on your specific case. Then I suppose that adding a proper message to hint about how to fix on known treacherous functions would be worth it. Also remind me of #8263 (it's the opposite, you should assign but you're not. Creating a generic message with a proper specific hint instead of a new message for each is better)

[orSolocate]

Hi @Pierre-Sassoulas , long time no see, nice to work with you again. I saw the #7853 comment, I could retry fixing the infer logic now that astroid is hopefully improved. But I think it can be done in parallel with implementing the issue here. A specific message for this one is needed in my opinion because, like in the list.sort() case, we want to advice the user to use the reversed() function. Could you enlighten how #8263 is related (it was hard for me to understand what it is about).

[Pierre-Sassoulas]

Hey likewise @orSolocate , glad to see you're back :)

I think the message can be generic with an added precision: i.e.:
Assigning result of 'foobar' to a variable, but 'foobar' returns None in the general case.
Assigning result of 'reverse' to a variable, but 'reverse' returns None, did you mean 'reversed(...)' ? for the specific case based on a dict:

KNOWN_TREACHEROUS_FUNCTIONS = {
    "reverse": "reversed",
    "sort": "sorted"
}

The same could be done for function that are known to be treacherous but that return something when you don't expect them to (the message in #8263 is kind of the opposite of assignment-from-none or assignment-from-no-return).