gh-87646: Added .path to tempfile

[ganesh-k13]

Added .path to generate pathlib's Path objects for NamedTemporaryFile and TemporaryDirectory

Note: Original suggestion was to make it a method so the Path creation will be lazy. I'm ok with both and wanted the opinion of a maintainer.

• Issue: Make tempfile.* classes path-like #87646

[AlexWaygood]

I like the idea; I'm always doing stuff like:

with TemporaryDirectory() as td:
    td_path = Path(td)

And it feels annoying!

I'd prefer it if we used functools.cached_property — I think it looks nicer as an attribute, but I think lazy evaluation is preferable here, or we'll be slowing down TemporaryDirectory() calls for users who aren't interested in this feature.

Also, this will need documentation to be added to Doc/library/tempfile.rst :)

[AlexWaygood]

Don't worry about test__xxsubinterpreters crashing; it's happening on main; see #104341

[ganesh-k13]

Thanks for the quick review! Yep that makes sense, let me make that change and add to docs 👍

[ganesh-k13]

I have pushed changes, comparison: https://github.com/python/cpython/compare/ee6435b52832cb6a102391bfddc92883170249f9..5d6a344082136c396ee3a47068cd047b48c8a59c.

I added a para on namedTemporaryFiles as it is not there today. Can remove/edit if it's not needed.

[barneygale]

I've left a comment on the issue about whether we should add a .path attribute or actually inherit from Path. I lean towards the latter at the mo.

[eamanu]

Shouldn't new path property be explained somewhere?

[ganesh-k13]

I thought same, but even the .name is not explained in this file. I can add it if someone can point an apt location 👍

[ganesh-k13]

NamedTemporaryFile indeed had a reference to .name, I added .path there. Thanks!, ref: diff. I'll leave the TemporaryDirectory for now, we may need to add name and path

[AlexWaygood]

(Nit: where possible, please avoid force pushes when contributing to CPython. They interact badly with the GitHub UI, making it hard to see what changed between commits. All PRs are squashed into a single commit when they are merged, anyway, so there's no need to worry about having a messy commit history on a PR :)

[ganesh-k13]

Ah ok. I wanted to know what was the convention here. Thanks a lot for letting me know! I'll follow this from now on.

[AlexWaygood]

Ah ok. I wanted to know what was the convention here. Thanks a lot for letting me know! I'll follow this from now on.

No need to read it all, but when in doubt, the CPython devguide has lots of information about CPython's specific workflow, which can be useful in these situations :) https://devguide.python.org

[ganesh-k13]

Hey @AlexWaygood / @barneygale, are any more changes needed for this PR?

[github-actions]

This PR is stale because it has been open for 30 days with no activity.

[serhiy-storchaka]

Let's not create unnecessary coupling. You always can call the pathlib.Path constructor or constructor of other path-like class if needed.