7.4.0 release plan

[AA-Turner]

I intend to release Sphinx 7.4.0 in the next few days.

7.4.0 will be the last version to support Python 3.9 -- I have the branch ready to drop support, but I don't see any great value in dropping before the release.

Please let me know if there are any features or bugfixes you would like to have included in the release, or any other matters.

I am currently looking to include:

• Require Docutils 0.20 or greater #12555
• LaTeX writer imposes too many constraints on desc_type_parameter_list nodes #12543 (PR at Fix PEP 695 output for classes in the LaTeX builder #12561)
• toc_object_entries uses language (Python) scoping instead of documentation scoping #12352 (PR at toctree, use document nesting instead of domain nesting when adding domain objects #12367)
• [LaTeX] Revamped styling of all admonitions, with addition of a title row with icon #12508
• LaTeX: for French language and lualatex engine polyglossia and not babel is used contrarily to Sphinx documentation #12410 (PR at [LaTeX] For 'lualatex' with French, use babel LaTeX package as default, as for 'xelatex', and not polyglossia. #12557)
• Potentially Support type-dependent search result highlighting via CSS #12474

A

cc: @sphinx-doc/developers @sphinx-doc/triagers

[picnixz]

I'll need a bit of time for #12508 to find the correct colors but I should be done by the end of the day

[jfbu]

Could you also consider #12557? (I added it right now)

[jfbu]

Regarding #12508 I will push some change to default border-widths and colours we were discussing with @picnixz.

Edit: I have pushed my changes. PR is ready from my side. @AA-Turner. But I had overlooked @picnixz comment above...

[AA-Turner]

For awareness and/or pre-release testing, cc also:

• @jbms / @2bndy5 (https://github.com/jbms/sphinx-immaterial/)
• @humitos (https://github.com/readthedocs/sphinx_rtd_theme)
• @agoose77 / @choldgraf (https://github.com/executablebooks/sphinx-book-theme)
• @pradyunsg (https://github.com/pradyunsg/furo)
• @drammock (https://github.com/pydata/pydata-sphinx-theme)
• @mgeier (https://github.com/mgeier/insipid-sphinx-theme/ & https://github.com/spatialaudio/nbsphinx/)

A

[mgeier]

The regression I reported in #12096 was at some point planned to be fixed for 7.3.0 (#12242) but it apparently didn't make it. It would be great if this could be fixed for 7.4.0.

[chrisjsewell]

I intend to release Sphinx 7.4.0 in the next few days.

If you could hold off on this for another ~24 hours, i.e. just until the end of Sunday, that would be appreciated cheers

[chrisjsewell]

Also, I don't expect it to make in 7.4, but it would be great to have your opinion on #12507 😄

[chrisjsewell]

Heya, @AA-Turner apart from discussing #12506 (comment), the other main thing I would like to look at getting in is #12567

---

I want to add descriptions to config values, using #12549, and finalise #12471, put probably I don't get around to that today

[jayaddison]

@wlach this isn't directly 7.4.0-related, but I noticed it during review of closed tickets, and discussion there is locked, so commenting here:

I closed #12045 as resolved by your PR #11958, that definitely does fix a bug in that part of the code. However: I think the iteration-over-all-terms still occurs, and given that we do store all terms in sorted order, I think that (potentially significant) performance optimization might still be possible. Let me know if you agree and if so I'll create another GitHub issue to handle that (since the old one can't be reopened while locked).

[jayaddison]

@wlach this isn't directly 7.4.0-related, but I noticed it during review of closed tickets, and discussion there is locked, so commenting here:

I closed #12045 as resolved by your PR #11958, that definitely does fix a bug in that part of the code. However: I think the iteration-over-all-terms still occurs, and given that we do store all terms in sorted order, I think that (potentially significant) performance optimization might still be possible. Let me know if you agree and if so I'll create another GitHub issue to handle that (since the old one can't be reopened while locked).

Hrm; an improvement here is not so obvious or straightforward after all. The partial-string matching allows query words to match within document terms as substrings -- not only by prefix as I thought. That means that we can't rely on binary-search on the sorted term list to find all possible matches.

Edit: attempt to clarify phrasing of match behaviour.