Add a regression tests for https://github.com/PyCQA/astroid/pull/1207

[Pierre-Sassoulas]

Type of Changes

	Type
✓	🐛 Bug fix
✓	📜 Docs

Description

This should fail before we upgrade astroid following the merge of pylint-dev/astroid#1207

[coveralls]

Pull Request Test Coverage Report for Build 1641041479

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 93.694%

---

Totals
Change from base Build 1641026660:	0.0%
Covered Lines:	14338
Relevant Lines:	15303

---

💛 - Coveralls

[Pierre-Sassoulas]

This does not fail like it should.

[DanielNoord]

Shouldn't we make this a blocker for 2.12.0? The change in astroid seems minimal and fixable before 2.12 is released!

[Pierre-Sassoulas]

I did not put the fix in astroid 2.8.5 because testing would take time. There are a lot of fixes in 2.12 (ex 2.11.2) I'd like to release it fast and cutting the scope will help. Also we'll do 2.13 very soon after because of python 3.6.2 deprecation.

[DanielNoord]

Ah okay! Would you mind waiting for this afternoon before releasing? I've got some undefined-variable issues I would like to fix as well, some of which already are with the new assignment expression logic. Think it would be nice to fix a bunch of them in this release!

[Pierre-Sassoulas]

Would you mind waiting for this afternoon before releasing?

There's still some hard to do issues like primer tests that I hesitate about moving in 2.13 so it's more like a "I'm trying to focus on releasing now" than "I'm going to release right now" at this point 😄

[DanielNoord]

@Pierre-Sassoulas I would be okay with letting this go. If anything this should probably be tested in astroid. This seems to be such a niche case that we can add tests cases when somebody reports problems with this (we will then probably also get more input on what to actually test).

[DanielNoord]

Note that this being tracked by #5341, so if we decide to close this that issue should be closed as well.

[Pierre-Sassoulas]

I'm reluctant to close as we merged in astroid with the implicit assumption that we were going to test in pylint as it would be easier to test in pylint. I totally lost the context of the original change at this point though.

[jacobtylerwalls]

I opened an alternative regression test that does fail before the astroid PR (and hopefully is meaningful!) in pylint-dev/astroid#1397.

This does not fail like it should.

I think this was expected, in the PR the author said "when you run the test build within such a no-filesystem environment."

[DanielNoord]

Superseded by pylint-dev/astroid#1397. Thanks a lot @jacobtylerwalls, this has been a very long open issue.