Cristian Le

@martinhoyer can you make a PR to bump the pre-commit. I think there would be some changes from the ruff linter

Doesn't it force it to multi-line if you manually add a comma at the end. We checked on this option somewhere in the history and there is an option for...

Well, most of the issues here is that a strict formatter was not enforced and this would deal with the downfall of that. I don't think there is a good...

Oh yeah, using the commit does indeed work :/

Agree with the idea, and I think it's not the first time this came up. One issue I can think regarding this is how it can interact with `provision.user`. I...

I've looked at a bit and I did not see an issue around user being overridden. But I've found another issue. All tests are performed in the same environment so...

> @LecrisUT I wonder if we really need to uninstall the packages? That would be to avoid unintentional side effects. Here's an example: I write 2 tests that test cmake...

Yes indeed with the current workflow that is also not possible. But it's partially about expectations. In the current approach it will always fail and you see the installation done...

> - keep this issue only for installation before running the test Maybe we could emphasize the usage of `order` if a full rollback is not necessary for each test.

> `Generic, Common, Base` doesn't tell anything about what parent class provides. > > If there are just some utility methods, I would prefer composition over inheritance. Interesting proposal. This...