What I can do is pin back theruff pre-commit to the same version / tag as ruff itself... though we'd then want to update it later when#960 or something similar goes in

However, the hook version doesn't seem to affect the version ofruff run.if I use--all-files, I do still see a couple errors withruff-lint, as well as onemypy error that doesn't seem to show up in CI. Either way, I think this will solve at least some of the problems with running it locally, but let me know if you want to pin back the pre-commit hook to match the ruff version.

However, the hook version doesn't seem to affect the version ofruff run.if I use--all-files, I do still see a couple errors withruff-lint, as well as onemypy error that doesn't seem to show up in CI. Either way, I think this will solve at least some of the problems with running it locally, but let me know if you want to pin back the pre-commit hook to match the ruff version.

I really believe this is because the pre-commit is not reading our config in the pyproject.toml file because if you see the lint job in the ci we have to specify the configuration file to ruff. Why it doesn't do this by default is besides me but it changes what is flagged in the CI

Also, just read that the lint should come before format if lint is run with—fix, so maybe I should switch the order?

Also, I can try adding that argument here and see if that fixes the issue.

Also, I can try adding that argument here and see if that fixes the issue.

So, just FWIW, I checked and adding--config=pyproject.toml to the args for pre-commit config doesn't seem to change the behavior (also, I noticed that ruff does seem to pick up settings frompyproject.toml even when it's not called with that argument). If it matters, I'm also not seeing codes for the specific findings (for example,ERA001) ignored in the config file.

Playing around with it locally, Ithink the actual issue is that having the--diff flag set seems to suppress some of the findings. If you tryruff check --fix vsruff check --fix --diff

$ ruff check --fix[...]tests/unit/semantic_release/hvcs/test_gitlab.py:  612:5 SIM117 Use a single`with` statement with multiple contexts instead of nested`with` statementstests/util.py:  15:34 N812 Lowercase`config` imported as non-lowercase`cliConfigModule`Found 26 errors.$ ruff check --diff$echo$?0

So even though there are findings,ruff check --diff seems to be exiting 0 in any case (note: I think it's not printing anything else because there are nofixable findings, even though there are findings).

That said, we can leave this as it is here, and I can try ignoring and / or chipping away some of those other findings in another PR.

Also, I think with this change, at least ruff should now only run against modified files.

I'm trying to fix the mypy pre-commit issue with--all-files too, but may have to figure that out separately. It basically has to do with the way files are provided / passed

Just settingsemantic-release/ as a finalargs value doesn't seem to do it properly, probably because it expects filenames to be passed in last as a positional argument, but withpre-commit run --all-files, it's throwing at least one error

% mypy semantic_release                                                         Success: no issues found in 46 source files% python3 -m mypy --config pyproject.toml --ignore-missing-imports semantic_releaseSuccess: no issues found in 46 source files% pre-commit run --all-files[...]semantic_release/cli/commands/publish.py:66:9: error: "HvcsBase" has noattribute "upload_dists"  [attr-defined]            hvcs_client.upload_dists(tag=tag, dist_glob=pattern)            ^~~~~~~~~~~~~~~~~~~~~~~~Found 1 error in 1 file (checked 46 source files)

Another thing I can do separately, but Ithink with the versions ofmypy andruff we have, we may not need to set the config file anymore for either?

Threw up a draft first pass in#961 at the rest of the ruff failures and removing--diff from the CI job

% python3 -m mypy --config pyproject.toml --ignore-missing-imports semantic_release
Success: no issues found in 46 source files
% pre-commit run --all-files
[...]
semantic_release/cli/commands/publish.py:66:9: error: "HvcsBase" has no
attribute "upload_dists" [attr-defined]
hvcs_client.upload_dists(tag=tag, dist_glob=pattern)
^~~~~~~~~~~~~~~~~~~~~~~~
Found 1 error in 1 file (checked 46 source files)

Let's just throw a# type: ignore on this because this will be refactored soon anyway.

Another thing I can do separately, but Ithink with the versions ofmypy andruff we have, we may not need to set the config file anymore for either?

Probably but not sure. The less command line arguments the better.

Probably but not sure. The less command line arguments the better.

Ok, see how you think this looks now - removed what I think should be unnecessary args, also added that ignore.

Thanks 🎉