Don't run flake8 twice in CI

[AlexWaygood]

We now run flake8, black and isort on PRs using pre-commit.ci. This is faster than using tox on GitHub Actions, and for black and isort, it also autofixes PRs (and the check in CI fails if it can't autofix a PR).

[hauntsaninja]

Hm good spot, but I'm not sure about this, since it's nice to have the guarantee that tox works. But don't feel strongly!

Edit: I guess we could remove flake8 from precommit, which is the slowest thing in the lint step

[AlexWaygood]

Hm good spot, but I'm not sure about this, since it's nice to have the guarantee that tox works. But don't feel strongly!

Yeah. I wondered if we should maybe delete the lint tox environment altogether. Contributors now have the following options to run the lint checks locally, which is probably too many options:

• Run them directly (black ., etc.)
• Setup pre-commit locally
• Run tox run -e lint
• flake8 is included in the checks done during runtests.py, and if Easy way to run lint, black etc. (all fast checks) #15076 is implemented, black and isort will also be added to runtests.py.

Edit: I guess we could remove flake8 from precommit, which is the slowest thing in the lint step

People who have pre-commit set up locally will probably want to run flake8 as part of pre-commit. But if we want it run separately, we could possibly skip it for CI runs of pre-commit, like we do at typeshed, and run it separately using GitHub Actions: https://github.com/python/typeshed/blob/9abf620500644509a0ba656767e858ac2ef23dd6/.pre-commit-config.yaml#L41-L46

[hauntsaninja]

Thank you!

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉