LINT - Run updated pre-commit hooks

[trallard]

Follows-up #2044

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
src/pydata_sphinx_theme
short_link.py
utils.py
Project Total

This report was generated by python-coverage-comment-action

[trallard]

Thanks for your help @Carreau these are some of the most annoying PRs 💜 I am very grateful for you

[drammock]

This approach to git-blame-ignore-revs doesn't work. See:

https://github.com/pydata/pydata-sphinx-theme/blame/main/docs/conf.py#L197-L198

The problem is that the PR was squash-merged, so the referenced commit hashes don't exist on main. When you're contending with pre-commit CI, the way I've seen it handled is to:

1. make a commit that only enables the new style rules and commit it with --no-verify
2. run pre-commit locally, then stage and commit the auto-changes
3. make a PR of those 2 commits, but then rebase-merge instead of squash-merge (needs to be temporarily enabled by repo admin in the settings)
4. make a new PR that adds the commit hashes to git-blame-ignore-revs

In principle it should be possible to do it all as one PR (with 3 commits: enable-the-rule, apply-the-rule, and add-hash-to-ignore) and then rebase-merge that PR. But I've seen such attempts go sideways before --- maybe the commit hashes changed when rebased onto main? don't recall the details. The 2-PR approach is much less error-prone.

[Carreau]

My bad, I forgot that the default here is squash-merge – I guess that's one of the reason I dislike squash-merge.

And yes the commit hash do change when rebasing as the hash depends on all the parents.

[trallard]

Right, so would you suggest we revert this merge @drammock, or shall we do something else to fix this (I suppose we could add the squashed commit to the .git-blame-ignore-revs file, but that has the downside of ignoring all the changes, including updates to the pre-commit config file?

Also it might be worth adding #2049 (comment) to our dev docs

[drammock]

I think what I would do is:

• revert f6102b4 and merge the reversion (done in 33333c8)
• open a new PR that re-creates f6102b4 but uses cherry-pick to split the changes into 2 commits: the automatic changes and the intentional ones (done in Redo precommit changes #2055)
• merge the PR from (2) using rebase-merge
• open a new PR adding f6102b4, the reversion commit, and the automatic-changes commit from (2) into to .git-blame-ignore-revs (done in update revs to ignore in blame #2056)

I can work on this now, and will try to preserve all the correct authorships.

[drammock]

All fixed! Now https://github.com/pydata/pydata-sphinx-theme/blame/main/docs/conf.py#L197-L198 points to #834 (2 years ago)