It seems like make suspicious is no longer used by Python releases: good.

• I cannot find suspicious in https://github.com/python/release-tools/
• I cannot find suspicious in https://peps.python.org/pep-0101/ neither

cc @pablogsal @Yhg1s

sphinx-lint is run by the "Docs / Docs" CI job on this PR: "Check documentation" step:

In short, sphinx-lint is now run on all documentation changes, not only late during releases. Issues are catched earlier: good!

Does this need an entry in the 3.12 whatsnew?

I don't think so as it's not a new thing, it's an old thing being cleared (it was a tool to help migration from LaTeX documentation to Sphinx documentation, so its removal is long overdue).

It's been one year that the CI no longer uses make suspicious, I bet everybody just forgot about it existence. And the few humans running make suspicious manually are getting a deprecated message since November 2021.

It's good that Doc/tools/susp-ignored.csv can be removed. It can be removed because sphinx-lint has less false positives, right?

sphinx-lint has 0 false positives on a growing list of watched repositories, as of today:

This list of repositories are automatically checked by sphinx-lint CI to "guarantee" no false positive are introduced as release time, like it happened a few days ago with the "default-role debacle" (which happened because this checker is not enabled by default, to it went under the radar, it's fixed now, we test each repos with their real set of checkers).

The goal is to stick to 0 false positive (aiming at the impossible... because why not).

I don't think so as it's not a new thing, it's an old thing being cleared (it was a tool to help migration from LaTeX documentation to Sphinx documentation, so its removal is long overdue).

The whatsnew also has a section for deprecated/removed APIs/modules/tools.

See for example

The whatsnew also has a section for deprecated/removed APIs/modules/tools.

OK, so, why not :)

The whatsnew also has a section for deprecated/removed APIs/modules/tools.

Done.

@JulienPalard, looks like the 3.11 and 3.10 branches still run make suspicious -- should this PR be backported?

@JulienPalard, looks like the 3.11 and 3.10 branches still run make suspicious -- should this PR be backported?

I don't think that it's a good idea to backport. It's better to leave these stable branches as they are.

That means that we will continue to get make suspicious false-positive errors on those branches, and we will continue to have to update Doc/tools/susp-ignored.csv for those branches when we get those false positives.

3.11 isn't running it anymore, since it got removed here:

• bpo-42238: Doc CI: Disable suspicious checks. #26575

3.10 however is still running it. If this is a problem we could just switch to sphinx-lint in 3.10 too, but the removal of make suspicious and rstlint.py shouldn't be backported.

3.11 isn't running it anymore, since it got removed here:

• bpo-42238: Doc CI: Disable suspicious checks. #26575

3.10 however is still running it. If this is a problem we could just switch to sphinx-lint in 3.10 too, but the removal of make suspicious and rstlint.py shouldn't be backported.

Thanks, that makes sense! I think I remember seeing some false positives recently on some backports — they must have been backports to 3.10.

Agreed that backports should also be as minimal as possible!

I propose just disabling make suspicious in 3.10 in #98221.

I don't feel the need to add sphinx-lint there: the PR will come from main, where they have already been checked, and as victor says, better not touch those stable branches too much (I mean let's remove pain, not add some).