Fix return type checkers calls on ellipses functions #5107
Conversation
DanielNoord
added
the
False Positive 🦟
Pull Request Test Coverage Report for Build 1312511792
💛 - Coveralls |
pylint/checkers/classes.py
Outdated
| @staticmethod | ||
| def _function_body_is_ellipsis(node: nodes.FunctionDef) -> bool: |
There was a problem hiding this comment.
What do you think about moving this method to checkers/utils.py?
There was a problem hiding this comment.
Shouldn't it be in astroid's FunctionDef instead ?
There was a problem hiding this comment.
Shouldn't it be in astroid's FunctionDef instead ?
Though about this some more. I'm not sure we should move these kinds of helper functions to astroid. astroid only really deals with the ast and inference, everything else should be done by pylint.
By the same logic I believe it might have been wrong to move is_sys_guard and is_typing_guard to astroid in the first place. They don't really belong there.
There was a problem hiding this comment.
Couldn't it be argued that these methods are part of the inference? In that they give additional information about the specific type of FunctionDef node?
Anyway, I'll leave this up to you and Pierre. I'm fine with either.
If wanted I can also make a PR to re-move is_sys_guard and is_typing_guard to pylint.
There was a problem hiding this comment.
Couldn't it be argued that these methods are part of the inference? In that they give additional information about the specific type of
FunctionDefnode?
True. Personally I would consider this a case of meta inference. We check multiple conditions to decide wether or not to emit a warning. That's that sort of thing pylint does all the time.
One thing to consider with that, it's much easier to iterate (read: fix) such a check in pylint alone than to modify something in astroid + pylint.
What I do consider useful in astroid are helpers like this one pylint-dev/astroid#1169 which helps dealing with the NodeNG tree itself.
Anyway, I'll leave this up to you and Pierre. I'm fine with either.
If wanted I can also make a PR to re-moveis_sys_guardandis_typing_guardtopylint.
I would be in favor of it, but let's see what @Pierre-Sassoulas thinks first. Unfortunately, we can't outright remove them now, so we would need to deprecate them first (and remove with astroid 3.0).
There was a problem hiding this comment.
Let's wait for this to release astroid 2.8.1 then. My local clone is broken and I need to fix it anyway.
There was a problem hiding this comment.
Do you mean first release 2.8.1 and afterwards deprecate, or the other way around?
There was a problem hiding this comment.
I guess, let's do the deprecation now. I.e. wait with releasing astroid 2.8.1 until it's done.
pylint-dev/astroid#1199
There was a problem hiding this comment.
I opened pylint-dev/astroid#1202
There was a problem hiding this comment.
Opened #5122 to add the guard helper functions back to pylint.
Co-authored-by: Marc Mueller <[email protected]>
| The "invalid-*-returned" messages shouldn't be emitted for stub functions | ||
| Original issue: https://github.com/PyCQA/pylint/issues/4736""" |
There was a problem hiding this comment.
Can you move these just before the class definition? Someone might add additional test cases later on. That way at least the general docstring will stay up to date.
|
Should be good to go now! |
doc/whatsnew/<current release.rst>.Type of Changes
Description
Closes #4736