Document 3.13, 3.14, 4.0 and future removals

[hugovk]

Follow on from #92308.

This lists in the What's New page some more deprecations pending removal in:

• 3.13 (#84540 - lib2to3 and 2to3, #93674)
• 3.14 (#93864)
• 4.0 (#57281)
• future versions (#83710)

[warsaw]

I wonder if there was some automation we can add to help ensure that planned deprecations actually happen? Like, maybe a JSON file or some such where the deprecations are recorded (or a script to parse the docs or code). Then, when the new version is created at main, the automation would open some release blockers on those deprecations. What do you think?

[arhadthedev]

@warsaw Probably, grepping docs for .. deprecated-removed:: and headers for Py_DEPRECATED would be better than maintaining another list file.

Edit: for \.\. deprecated-removed:: [^ ]+ (\d+).(\d+) and Py_DEPRECATED\((\d+).(\d+)\).

[hugovk]

I wonder if there was some automation we can add to help ensure that planned deprecations actually happen?

Yes, using the new warnings._deprecated method will make it clear with RuntimeErrors after the relevant alpha: https://discuss.python.org/t/introducing-warnings-deprecated/14856

[ezio-melotti]

I wonder if there was some automation we can add to help ensure that planned deprecations actually happen?

See also:

• Enforce the use of deprecated-removed in docs #92564
• Add support for future-version workflows core-workflow#468

[ezio-melotti]

I left a comment and a couple of minor suggestions -- feel free to disregard them if you think the current version is better.

[CAM-Gerlach]

I wonder if there was some automation we can add to help ensure that planned deprecations actually happen?

You could also fire a Sphinx warning in the else branch here in the directive (with a check for whether its an alpha, and perhaps an :option: to disable it for special cases), just like warnings._deprecated does.

[bdarnell]

The deprecation plans for this function have changed. See #100970 for the updated deprecation notice.

[hugovk]

Thanks for the note! This PR has got pretty out of date...

[hugovk]

Will try again in a new PR :)

[hugovk]

Please see PR #108055 instead 🚀