Backport PR #46558: CI: pre-commit autoupdate to fix CI

[MarcoGorelli]

Backport PR #46558

[simonjayhawkins]

@MarcoGorelli There are a few CI PRs to manually backport. starting with #46543 (have opened #46559)

It maybe better to not merge this one yet. (makes autobackports more likely to succeed)

Out of curiosity, where were the conflicts that prevented the autobackport of #46558 succeeding

[MarcoGorelli]

sure, will hold off on this one

they were conflicts with another PR which updated black version pins

[simonjayhawkins]

Thanks. Have marked as draft and will do the backports in order, so this one after #46541 and #46540

[simonjayhawkins]

they were conflicts with another PR which updated black version pins

I guess #46490 which we didn't backport. so probably autobackport won't suceed even after getting all the ci stuff in sync.

[simonjayhawkins]

code check are failing, I assume since we did not backport #45752

in #46558 (comment), I said ...

I assume will will get the same error on 1.4.x?

even though we are not seeing the same failures on 1.4.x

It maybe that since black on 1.4.x is pinned to 21.12b0 whereas on main it was 22.1.0 that we don't need the backport.

Alternatively, since the black changes should not affect code behavior, we could run the latest black as a commit here (I assume that would be easier than backporting #45752 and #46490)

[MarcoGorelli]

Looks like it works fine in 1.4.x - maybe 21.12b0 isn't affected by the latest click release then?

I'd suggest to then either not backport black changes at all, or if click does cause issues, then just to pin that in additional_requirements for this hook

Or will not backporting black changes make other backports tricker?

[simonjayhawkins]

Or will not backporting black changes make other backports tricker?

There we many files changed in #45752 so there is perhaps a greater risk that a manual backport is required in the future due to a conflict.

I'm +1 on updating the code on 1.4.x to the latest black (although IIRC we have not tended to do this in the past)

[MarcoGorelli]

sure, I've done this in #46576, will close this one