Top-level code environment docs are outdated for Python 3.13+

[wimglenn]

Bug report

Bug description:

Ref: https://docs.python.org/3/library/__main__.html

There is a script and entry-point example which has a conditional upon __main__.__file__ and also a remark like:

Note that in this case the __main__ scope doesn’t contain a __file__ attribute as it’s interactive.

This was all working as demonstrated up to and including Python 3.12, but starting in Python 3.13 it's incorrect. The attribute now exists, and is pointing at _pyrepl/__main__.py.

The conditional demonstrated in namely.py:print_user_name now always goes into the if branch, and no longer the else branch, resulting in confusing output.

Ping also @jdevries3133 who wrote the original prose in #26883

CPython versions tested on:

3.13

Operating systems tested on:

Linux

Linked PRs

• gh-128720: Fix doc inacurracy about __main__.__file__ existence in the REPL #128721

[jdevries3133]

Oh no! Thanks for pointing this out. Would be happy to jump in and send a PR to update the docs. If anyone has context to share about whether or not the new behavior is intended, that would be great!

I'll start taking a more detailed look in the meantime!

[AA-Turner]

If anyone has context to share about whether or not the new behavior is intended, that would be great!

I think __file__ being defined in the interpreted is surprising and likely unintentional, though perhaps @lysnikolaou or @pablogsal would be able to comment.

A

[pablogsal]

I think __file__ being defined in the interpreted is surprising and likely unintentional, though perhaps @lysnikolaou or @pablogsal would be able to comment.

It is not intentional but most of these alterations are due to pyrepl being a Python module. The documented behavior is prescriptive of the previous implementation more than an assurance that should be maintained IMHO. On the other hand assuming that nothing breaks, I am happy to take a PR that removes the attribute from the module to preserve the documentation :)

[wimglenn]

My 2c: The attribute is correct and should stay. Why hide the fact that the pyrepl is a Python module?

Unlike sys.ps1, there wasn't really a documented guarantee on __file__ presence/absence in the REPL.

The page in the linked PR would be more focused and simpler without this extra logical branch. It's a helpful page, but the conditional on the __file__ attribute in the REPL seems tangential to the main point.

[pablogsal]

I agree! Let's go with #128721 (review)

Thanks a lot for the PR @wimglenn