gh-129393: Make 'sys.platform' return "freebsd" only on FreeBSD witho…

[michael-o]

…ut major version

Result:

osipovmi@deblndw011x:~/var/Projekte/cpython (main *+>)
$ /tmp/python-3.13/bin/python3
Python 3.14.0a4+ (heads/main-dirty:8e57877e3f4, Jan 28 2025, 10:01:48) [Clang 19.1.5 (https://github.com/llvm/llvm-project.git llvmorg-19.1.5-0-gab4b5a on freebsd
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.platform
'freebsd'
>>>

• Issue: Make 'sys.platform' return "freebsd" only on FreeBSD without major version #129393

---

📚 Documentation preview 📚: https://cpython-previews--129394.org.readthedocs.build/

[cpython-cla-bot]

All commit authors signed the Contributor License Agreement.

[michael-o]

FWIW: I am a FreeBSD committer as well.

@emaste

[encukou]

cc @vstinner as the FreeBSD contact

[emaste]

LGTM. Hopefully there's not too much fallout.

[michael-o]

LGTM. Hopefully there's not too much fallout.

Me as well, but this should stay on 3.14 and not be backported, even if I'd like to see that.

[vstinner]

LGTM

[vstinner]

IMO this change is backward compatible since it was recommended to use sys.platform.startswith('freebsd') to check if Python is running on FreeBSD. This code remains correct is sys.platform becomes "freebsd".

[vstinner]

We did the same change on Linux when Linux was upgrade from Linux 2.x to Linux 3.x, and it went fine.

[michael-o]

IMO this change is backward compatible since it was recommended to use sys.platform.startswith('freebsd') to check if Python is running on FreeBSD. This code remains correct is sys.platform becomes "freebsd".

Oh great, I'd love to see this backported!

[vstinner]

Oh great, I'd love to see this backported!

That would be backward incompatible, it's a bad idea. This change can only land into Python 3.14.

[michael-o]

Oh great, I'd love to see this backported!

That would be backward incompatible, it's a bad idea. This change can only land into Python 3.14.

I am confused, doesn't this contract your previous statement: #129394 (comment)?

[emaste]

Code "should" be checking startswith but I'd wager there are there cases that use the full string; they shouldn't be broken in an existing release.

[michael-o]

Code "should" be checking startswith but I'd wager there are there cases that use the full string; they shouldn't be broken in an existing release.

That's correct and this is what I have noticed with pyproject.toml.

[michael-o]

Is there anything required from me to merge this one?

[vstinner]

Merged, thank you @michael-o.

[michael-o]

Merged, thank you @michael-o.

Thank you very much for the quick turnaround. Always a pleasure to work with pros!