Resolves junction links when deleting packages installed as editable

[jiunbae]

Resolve #10696

Or would it be better to use pathlib.Path?

[uranusjr]

Or would it be better to use pathlib.Path?

You can if you prefer, but this is also fine. I think we want to check both the original recorded path and the real path though (and pass if either of them match).

[jiunbae]

I don't know perfectly how pypi pakcages are stored and managed, so I thought it would be sufficient to make sure that the absolute path of the package path and the absolute path of the saved path are the same when deleting.

[uranusjr]

Let’s call realpath on both sides them? (Or use os.path.samefile?)

[jiunbae]

Does this look good to you?

assert os.path.samefile(link_pointer, os.path.realpath(dist_location)), (
    f"Egg-link {link_pointer} does not match installed location of "
    f"{dist.raw_name} (at {dist_location})"
)

or re-define dist_location as realpath?

[uranusjr]

I think samfile actually calls realpath underneath so we don’t even need that.

Edit: Oh, not actually, it’s using stat instead. But I think the Windows API should resolve junctions for stat. Can you try removing realpath and relying on only samefile and see if that works?

[jiunbae]

I checked that it works well in the environment presented in the #10696 . The os.path.samefile seems to work as expected in terms of operation.

[sio]

Using os.path.samefile should fix a similar failure I've encountered due to shortened Windo~1 paths:

AssertionError: Egg-link c:\users\runner~1\appdata\local\temp\makefile.venv_test_7o0hpfn3 
does not match installed location of hello 
(at c:\users\runneradmin\appdata\local\temp\makefile.venv_test_7o0hpfn3)

Is there anything that blocks this PR from being merged? @uranusjr?