jakkdl
commented
1 year ago IMO we could drop the flake8 checks immediately, if there's nothing else left out?
I don't mind running both for a while to see where they agree/disagree. But if it becomes bothersome I'll quickly start disabling error codes/plugins, set the stage for flake8 to manual, and/or remove it.
If enabling auto-updating of pre-commit, flake8-eradicate will come to bite us as it's not accepting PR's to show that it works with flake8 6. I was planning to add an updated fork as additional_dependency to flake8 instead, but I stumbled upon https://github.com/charliermarsh/ruff instead - which, among others, reimplements flake8-eradicate. Curiosity peaked, I added it to CI, enabled everything (it sucks there's no
--select-all), and after playing around for a bit it looks to be almost strictly better than flake8 in a lot of ways. It implements a ton of plugins by default, has autofix for a bunch of them, and in general very easy to transition to from flake8.flake8-mutable is the only plugin flake8-trio uses I didn't see listed in https://github.com/charliermarsh/ruff#rules (though maybe there is a rule for default mutable arguments anyway, haven't checked), and flake8-noqa (but there is support for checking noqa stuff). Looks like they haven't kept up reimplementing all rules from flake8-bugbear (esp not from the opinionated ones), but majority of them are in.
So might be interesting to enable it, and see how viable it is as a replacement for flake8.