gh-132185: Avoid slow test_expanduser_pwd2 in test_posixpath

[itamaro]

On systems where pwd.getpwall() can return lots of entries, this test can end up being very slow, for no obvious benefit.

• Issue: test_expanduser_pwd2 in test_posixpath can be very slow #132185

[sobolevn]

Just for the historic reasons: when we say "a very large number of entries" - what number do we mean exactly? 100? 100_000?

[itamaro]

Just for the historic reasons: when we say "a very large number of entries" - what number do we mean exactly? 100? 100_000?

I can't disclose the exact number since it's from an internal environment, but I can say the "multi-minute run time" is for an output size that's O(100k)

[serhiy-storchaka]

See also #132231.

[itamaro]

See also #132231.

nice!
could we do the "choose only 100" regardless of whether the CPU resource is provided or not?
I understand that it's "acceptable" to do this if the CPU resource is provided, but I'm not sure what benefit we're getting from it, considering the "choose 100" thing should be exercising the functionality we want to test.