I'm not convinced that it's a good idea to have os.statx_result which can be used instead of os.stat_result. It's not exactly the same, os.statx_result members can be None.

It's also strange that some statx members have a st_ prefix, whereas others have a stx_ prefix. I read the statx manual page and expect to have only stx_ prefix: same members names.

I don't think that stx_ prefix would prevent using statx() in pathlib in the future.

cc @jbosboom @cmaloney

I updated my PR to document all statx_result members, and I renamed stx_birthtime to stx_btime.

I changed my mind and just removed the comparison to stat_result:

statx_result has all the attributes that stat_result has on Linux, but using stx_ prefix instead of st_. statx_result is not a subclass of stat_result and cannot be used as a tuple.

I don't think that it's useful to compare these two structures, they are different and listing differences is not worth it.

I think this needlessly complicates code that uses os.statx on Python >= 3.15 and falls back to os.stat on earlier versions, but I guess there won't be any user code doing that if we get the pathlib interface in 3.15 as well, so +0 from me.

@vstinner You've made substantial changes to both the interface and implementation of os.statx. I think you ought to add your name in the blurb for the issue.

@vstinner You've made substantial changes to both the interface and implementation of os.statx. I think you ought to add your name in the blurb for the issue.

Ok, I added myself to the credits.