Fix crash in super-init-not-called checker

[DanielNoord]

• Add yourself to CONTRIBUTORS if you are a new contributor.
• Add a ChangeLog entry describing what your PR does.
• If it's a new feature, or an important bug fix, add a What's New entry in
  doc/whatsnew/<current release.rst>.
• Write a good description on what the PR does.

Type of Changes

	Type
✓	🐛 Bug fix

Description

Closes #6027.

Should have never put that comment there...

[coveralls]

Pull Request Test Coverage Report for Build 2069291873

• 6 of 6 (100.0%) changed or added relevant lines in 1 file are covered.
• 23 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.003%) to 94.215%

---

Files with Coverage Reduction	New Missed Lines	%
pylint/checkers/variables.py	23	96.51%

Totals
Change from base Build 2067203898:	0.003%
Covered Lines:	15439
Relevant Lines:	16387

---

💛 - Coveralls

[DanielNoord]

Uhm.. Does pypy support ctypes?

[Pierre-Sassoulas]

Maybe we need to forbid the use of infer accross all the pylint codebase, there's no way we're not going to get a crash eventuellay with the diversity of code pylint is going to be run on.

[Pierre-Sassoulas]

We might need to do

[testoptions]
except_implementations=PyPy

What do you think @DanielNoord ?

[DanielNoord]

We might need to do

[testoptions]
except_implementations=PyPy

What do you think @DanielNoord ?

I mean we could, but I don't know why this doesn't work on PyPy. I don't know the ctypes module...

[jacobtylerwalls]

Looks like the PyPy behavior is correct -- super's init is not being called, so the test file should have a # [super-init-not-called], but because astroid was raising InferenceError, it wasn't working for CPython -- which is good for our present purposes -- we want to test that no crash happens on an inference failure. So maybe a second file with the correct message for PyPy would do it?

[DanielNoord]

I guess raising if we can't infer will lead to more issues right?

[Pierre-Sassoulas]

I guess raising if we can't infer will lead to more issues right?

Yes I think we should prefer a false negative always when we're not sure.