chore: make sure local ruff runs don't touch vendored

[henryiii]

It's useful to run ruff locally sometimes, like to add --unsafe-fixes. This ensures it doesn't format or change something it shouldn't. See https://github.com/pypa/wheel/pull/617.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 71.03%. Comparing base (78b9ea9) to head (74c0d54). Report is 7 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #618   +/-   ##
=======================================
  Coverage   71.03%   71.03%           
=======================================
  Files          13       13           
  Lines        1084     1084           
=======================================
  Hits          770      770           
  Misses        314      314           

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[DimitriPapadopoulos]

I suggest you remove this: https://github.com/pypa/wheel/blob/1e00742acc9fb33f6e71460c3844c2b66532af7f/pyproject.toml#L71-L74

It had been forgotten when shifting from black to ruff in https://github.com/pypa/wheel/commit/83b77e591e6f593470f8daddf3bcfd6ca64e81bc.

[agronholm]

We already have this in place which makes this PR redundant.

[DimitriPapadopoulos]

Nope. it's not redundant. It's set for pre-commit, but not for standalone ruff runs.

[agronholm]

Why not just do pre-commit run -a?

[DimitriPapadopoulos]

Because I simply run ruff.

[agronholm]

Well, you could just as easily run pre-commit run -a ruff, no need to even install ruff explicitly?

[DimitriPapadopoulos]

Yet, I do have ruff installed. When I run ruff, results are not consistent with pre-commit.

[agronholm]

How so?

[agronholm]

Are you sure you're running the same version in both cases?

[DimitriPapadopoulos]

Of course. It's just that pre-commit skips vendored, while standalone ruff does not.

[henryiii]

It’s hard to add flags when running pre-commit, like --unsafe-fixes, so it’s handy to run locally sometimes. Unlike flake8, there’s no variations based on what else is installed, so it’s safe to run it outside pre-commit’s venvs.

Though if all you want to do is run it only with no special flags, I would recommend pre-commit run -a ruff.

[agronholm]

Of course. It's just that pre-commit skips vendored, while standalone ruff does not.

Oh, that's the only thing you meant by results not being consistent?

[agronholm]

It’s hard to add flags when running pre-commit, like --unsafe-fixes, so it’s handy to run locally sometimes. Unlike flake8, there’s no variations based on what else is installed, so it’s safe to run it outside pre-commit’s venvs.

Though if all you want to do is run it only with no special flags, I would recommend pre-commit run -a ruff.

Ah...that's a valid argument. Yeah, I don't see a way to pass flags through pre-commit run. When I needed to use --unsafe-fixes, I just temporarily added that flag to .pre-commit-config.yaml.

[DimitriPapadopoulos]

I tend to avoid pre-commit, it doesn't work well outside containers and virtualenvs on older Linux distributions, typically the oldest still maintained Ubuntu LTS.