Fix regression on ClassDef inference

[DanielNoord]

Steps

• For new features or bug fixes, add a ChangeLog entry describing what your PR does.
• Write a good description on what the PR does.

Description

Type of Changes

	Type
✓	🐛 Bug fix

Related Issue

Closes pylint-dev/pylint#5030
Closes pylint-dev/pylint#5036

[DanielNoord]

I'm not sure about the location of the unit test. Might need to be placed in a different place, but I did not know where.

[cdce8p]

node.keywords should never be None in the first place. It should always be a list, even if empty.
This should be fixed in ClassDef.postinit with

if keywords is not None:
    self.keywords = keywords

https://github.com/PyCQA/astroid/blob/28411ee8575562b7d7d1d0639d4e61f91ed1586b/astroid/nodes/scoped_nodes.py#L2099

--

ClassDef(identifier name,
             expr* bases,
             keyword* keywords,
             stmt* body,
             expr* decorator_list)

https://docs.python.org/3.10/library/ast.html

[DanielNoord]

Before merging this, do we want to identify why this regressed? Or is fixing the postinit good enough in your opinion?

[cdce8p]

The fix is good enough for me. Did some digging nevertheless.

That was an existing issue that only with the last update started causing problems. #979 added inference for Compare nodes. The inference method uses a clever workaround to get to the actual value of a node.
https://github.com/PyCQA/astroid/blob/28411ee8575562b7d7d1d0639d4e61f91ed1586b/astroid/inference.py#L811-L814

Calling as_string on any NodeNG should be safe. Multiple ClassDef.postinit calls in astroid/brain/brain_typing.py don't pass a keywords argument and thus use the default None.
https://github.com/PyCQA/astroid/blob/28411ee8575562b7d7d1d0639d4e61f91ed1586b/astroid/brain/brain_typing.py#L350-L354

I wouldn't add keywords=[] to these calls as a default value is expected to work fine. Thus this was an error in the postinit method which is now fixed.

[DanielNoord]

Thanks for the deep-dive! Feels good to know that we are not using a work-around to fix a problem but rather a fixing an actual problem, which was what I was afraid of!

[cdce8p]

I agree

[cdce8p]

LGTM!