GNU testsuite comparison:

Skipping an intermittent issue tests/timeout/timeout (passes in this run but fails in the 'main' branch)

Could you please add a test to make sure we don't regress? Thanks

Thank you for your feedback. I'll add a test.

In the issue #7285, when doing cp -v -i, the file is diff'd by timestamp and --update=older is used. The same test would use filetime crate, but with the following target_os limitation:

#[cfg(any(target_os = "linux", target_os = "android"))]
use filetime::FileTime;

Therefore, in the added test, cp -v -i is performed without --update=older and its output is checked.

Therefore, in the added test, cp -v -i is performed without --update=older and its output is checked.

This test verifies the existing behavior, which is fine, but it doesn't account for the changes you made in cp.rs, as it passes without those changes.

Additionally, I think you don't need filetime::FileTime and can use File::set_modified instead (it's used in some other cp tests).

Thank you for your feedback. My understanding was that the tests were lacking checks for output order in my changes. I've updated the tests in a new commit accordingly. If my understanding is incorrect, please let me know.

I think there is some confusion. Let me clarify: your change is about fixing the behavior of cp -v -i --update=older new old from

$ cp -v -i --update=older new old
'new' -> 'old'
cp: overwrite 'old'?

to

$ cp -v -i --update=older new old
cp: overwrite 'old'? y
'new' -> 'old'

and

$ cp -v -i --update=older new old
cp: overwrite 'old'? n

That works fine. However, your test doesn't ensure this new behavior works. It ensures cp -v -i new old works. And that's not the same as cp -v -i --update=older new old.

Thank you for your feedback and clarifications. I have implemented the changes accordingly; could you please review them?
I now recognize that my understanding of the --update=older option was lacking.

Thanks!