Use flake8 in pre-commit checks

mentioned in merge request !1441 (merged)

assigned to @fwinkl

Good idea. In ATLAS we run it as part of the build, but for Gaudi this seems the best solution given we already have other pre-commit checks. Will iterate on the exact list of error codes once the dust has settled after the open MRs are merged.

What about going for ruff? It claims to be a near drop-in replacement for flake8 with some additional benefits. The main selling point is the speed (not sure if that's an issue for Gaudi, though). However, there are some other nice aspects such as no need for plugins, a replacement for isort and other tools, compatibility with black, integration with pre-commit and VSCode, etc.

I'm not particularly attached to this tool (and I know the latest release at the time of writing is 0.0.263), and I don't mean to start a war. I just thought that since the issue came up only now, it's worth considering alternatives, especially since ruff seems to be getting momentum in the Python community with some big users jumping on the board.

I am not attached in particular to flake8 apart from the fact that we have lots of experience in ATLAS and I have zero on ruff (looks nice though). We also developed some ATLAS-specific flake8 plugins.

Anyhow, the main goal here is to get and keep our python code clean. In order not to delay this further I would suggest to go ahead with flake8 now and if someone is willing we can look at alternatives later.

Sure, I agree!

@cburr mentioned it to me too. It is being used in DIRAC and LHCbDIRAC and there it showed to be orders of magnitude faster.

In Gaudi we do not have that many Python files, so I think flake8 is already good enough.

Since @fwinkl did all the work and it's already used in ATLAS, I'm keen on merging !1441 (merged) as is and, maybe, switch to ruff at a later stage.

closed with merge request !1441 (merged)

mentioned in commit 8cfb95e6

mentioned in commit fb35755b

Use flake8 in pre-commit checks

Child items 0

Activity