gh-112451: Prohibit subclassing of datetime.timezone.

[felixxm]

This is consistent with C-extension datetime.timezone.

• Issue: C-extension datetime.timezone is not acceptable as a base class; pure-Python version is #112451

[erlend-aasland]

This will need a deprecation period as per PEP-387.

[felixxm]

This will need a deprecation period as per PEP-387.

Great, my first deprecation in Python 🎉 WIP

[felixxm]

Updated 👍

[pganssle]

This will need a deprecation period as per PEP-387.

I guess it isn't the worst thing in the world to have this deprecation period, but I think for PEP 399 compliance, especially when the feature is available only in the pure-Python module (which is not commonly used), there's a case to be made that this doesn't need a deprecation period.

I'm sure I'll eat my words when some PyPy-only user shows up and tells us they have been subclassing datetime.timezone, but I feel like it wouldn't be a huge deal if we just removed this functionality in the pure Python module? I don't want this question to block this PR indefinitely, so if it takes a long time to get a response we should go ahead with the warning, but I opened an issue with the steering council for clarification on this question.

[erlend-aasland]

Oh, TIL PEP-399; thanks, Paul. I'm always fascinated on the various ways people depend on very specific behaviour. Thanks for opening an issue over at the SC tracker.

[felixxm]

I reverted the deprecation per the Steering Council decision.

[pganssle]

Looks good to me, thanks for bearing through the uncertainty @felixxm !

[erlend-aasland]

Yes, thanks @felixxm! And thanks a lot to @pganssle and @serhiy-storchaka for your guidance 🙏

[felixxm]

Thanks y'all for help and reviews 🥇