Mazine Mrini

Yep, that's exactly what we currently are doing. > What do you actually expect here? Nothing much really. It just is an issue to track the typing as it is...

> There are still some cases that might not be correct: @radoering I thought `0.2-alpha1` was not a valid version, but looking through PEP440, it is. I will add the...

My last comment feels a bit too much opinionated so I kept the precision in `next_breaking()`, but maybe, `next_breaking()` should live in `PEP440Version` as it keeps precision

Here are some example of what `stable` used to do: ``` version => stable 0.2 => 0.2 0.2-a1 => 0.2.1 (adds precision) 1 => 1 1-alpha.1 => 1.0.1 (adds precision)...

> We're talking past each other. I've no problem with `stable()` being fixed to preserve precision. I am (i) asking why it moved to a different class and (ii) suggesting...

> your tests for `stable()` will then want moving (because they go with the `Version` and not the `PEP440Version`. And I think that will expose some differences eg `1.2.3.4dev0.stable()` will...

@radoering would you be so kind to rereview this PR and there seems to be an issue with SonarCloud which thinks the version `1.2.3.4` is an IP address

:wave: Just adding on top of this, `>0.0.2-alpha.21` yields `>0.0.2-alpha.21,

@dimbleby Opened https://github.com/python-poetry/poetry-core/pull/475

@dimbleby Looking at the PR, is that really the problem ? The issue I had was: We have a package with `sqlalchemy2-stubs=^0.0.2-alpha.21` and when building the wheel, it resolved into...