gh-66646: Explain __base__ attribute in the docs

[furkanonder]

• Issue: __base__ undocumented #66646

Co-authored by: Anupama Srinivas Murthy

[merwok]

I’m not sure that this implementation detail should be documented in three places.
One place with an index should be enough. Please also use the right markup to mark it as an impl. detail.

[furkanonder]

@merwok Since class.__bases__ is defined in stdtypes.rst and datamodel.rst files, I wanted to define class.__base__ in these files.

[merwok]

I understand your reason, but still think one place would be better than 3. I would like to see other people’s opinion!

For the markup, I meant this: https://devguide.python.org/documentation/markup/#codecell33 🙂

[furkanonder]

I understand your reason, but still think one place would be better than 3. I would like to see other people’s opinion!

CC: @hugovk @CAM-Gerlach

[hugovk]

I don't have a strong opinion, but if it's an implementation detail, fewer places might be better. The search should help people find it when they're looking.

[erlend-aasland]

@furkanonder, can you please rework this PR to align with the comments Raymond, Guido, and Éric gave on the issue?

[furkanonder]

@erlend-aasland Based on Raymond, Guido, and Éric's feedback, I believe changes have been made. Could you please let me know if there is anything I missed?

[erlend-aasland]

@erlend-aasland Based on Raymond, Guido, and Éric's feedback, I believe changes have been made. Could you please let me know if there is anything I missed?

Not at all. I will repeat their comments here:

Raymond said in #66646 (comment):

I think that base was intended to be a non-guaranteed implementation detail for new-style classes just like dictoffset and flags .

Guido followed up in #66646 (comment):

Raymond is correct. base is closely tied to the C implementation. I think it might be worth a mention, but with an explicit note that it is CPython-specific. It is sometimes useful, e.g. when wondering why a certain case of multiple inheritance is rejected.

So, we're talking about a non-guaranteed implementation detail. You've also been told multiple times that a single mention is sufficient, yet you do not follow up this; your PR still changes three doc files, including the reference. "Worth a mention" is far from your PR which adds 64 (!) lines.

I'm sorry, but I cannot see that you've taken any review or issue remarks into account.

[encukou]

@erlend-aasland, do you want to request more changes?

[willingc]

Thanks for rebooting this @encukou.

[merwok]

I disagree a bit with the label type-feature: shouldn’t we backport this small but maybe useful doc addition?

(skip news is fine)

[encukou]

Yeah, we generally backport docs improvements.

[miss-islington-app]

Thanks @furkanonder for the PR, and @encukou for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13, 3.14.
🐍🍒⛏🤖

[bedevere-app]

GH-140365 is a backport of this pull request to the 3.14 branch.

[bedevere-app]

GH-140366 is a backport of this pull request to the 3.13 branch.

[merwok]

Thanks to all involved!