QOL: easy way to run lint locally that matches the behavior in CI

[vkuzo]

We have the CI lint rules here: https://github.com/pytorch/ao/blob/main/.github/workflows/ruff_linter.yml , currently expressed as

        ruff check .
        # --isolated is used to skip the allowlist at all so this applies to all files
        # please be careful when using this large changes means everyone needs to rebase
        ruff check --isolated --select F821,F823,W191
        ruff check --select F,I
        ruff format --check || {
          echo "Ruff check failed, please try again after running 'ruff format'."
          exit 1
        }

It would be nice to have this logic be reusable, so I can run it locally with something like ./lint_my_code.sh and know that CI will be green. I used to do this with ruff format ., but that no longer matches the CI logic, slowing down the iteration cycle on PRs that pass ruff format . but fail the more extensive checks.

[jerryzh168]

yeah I already added it: https://github.com/pytorch/ao/blob/main/scripts/run_ruff_fix.sh

does this work for you?

[vkuzo]

can we make both CI and that script use the same rules, so the chance of divergence from future PRs is reduced?

[jerryzh168]

can we make both CI and that script use the same rules, so the chance of divergence from future PRs is reduced?

yeah I also commented with the same before, although I looked again and CI is doing check but the script is doing fix, we can have a arg to say we are checking or fixing though, I'll submit a PR

[jerryzh168]

Added here: #1450

[drisspg]

Updated here: #1453
Left comment about making sure to keep isolated checks in sync.

For more context the way to do this is to use pre-commit run. I would highly recommend people just install these hooks

pre-commit install they are very lightweight