Sven Panne

What exactly did https://github.com/PyCQA/astroid/commit/283befa92865faad7847b2d40a569e03ae00f7cc#diff-82f7976cc16d2319c09d26ba124fc46b fix? The previous behavior looked more correct to me, at least it didn't break the (very valuable) NewType. Can't we simply revert that "fix"?

Looks like a duplicate of #2296. Are there any plans of fixing this soon? This is a real show-stopper for pylint in projects which take typing a bit more seriously.

Well, my proposal is simply to revert the commit mentioned in #2296, as it obviously does more harm than good...

I am not really sure if this bug should be closed: Using `MutableMapping` instead of `Mapping` seems to be wrong at several other places, too: ` _TextMapping`, which is mutable,...

It's mainly because we pass `Mapping`s as parameters for cookies/headers/proxies IIRC, I would need to take a closer look if we make use of covariance in the values. As a...

I don't think that this is a balancing act, regarding typing things are always crystal clear: If a parameter gets mutated by a function/method it _must_ be a `MutableMapping`, using...

> When "balancing", we usually consider false negatives (no errors for wrong code) less bad than false positives (errors for correct code). I am not exactly sure who "we" is,...

> If people are still getting what they believe to be false-positive errors following #7696, it would be helpful if they could post a minimal repro and the exact error...

I think this issue should be reopened, the bug is still present in a relatively recent pipenv: ``` $ pipenv --version pipenv, version 2023.9.1 $ pipenv install requests Creating a...

Just a quick follow-up: The latest pipenv version (2023.10.24) has the bug, too, same repro as above.