Posix1e funcs in FreeBSD
Robert Watson
rwatson at FreeBSD.org
Thu Apr 24 06:58:32 UTC 2008
On Thu, 24 Apr 2008, Andrew Pantyukhin wrote:
> When looking at our ACLs, and how 3d-party code might use them (e.g. python
> libacl for ACL support in rdiff-backup), I see there are missing parts. For
> example, acl_copy_ext/int, acl_size are missing. These three are present in
> Linux.
>
> My question is, what's the status of posix1e-related work in FreeBSD and
> TrustedBSD. Is it intentionally slowed down, or will new work be accepted?
> If the latter, what's the right place to send patches to and discuss them?
My recollection is a bit hazy, as it's been several years, but I believe that
we decided not to implement those interfaces on the basis that the standard
didn't say what the external storage format was. As a result, there was no
portable way to implement the functions. Instead, we recommend using
acl_from_text(3) and acl_to_text(3), which do have defined formats. If you
are interested in implementing those calls, my initial recommendation would be
to make them wrappers around the text interfaces, although it would be worth
looking at how other systems do it (warning: other systems may not all do it
the same way).
The reason ACL work has slowed down (virtually stopped) is that it's
essentially done. There are probably a few bits and pieces in the system that
would benefit from better ACL support -- for example, I have a sneaking
suspicion that one or two utilities fail to properly preserve ACLs (nvi?).
The most recent parts to gain proper ACL support were tar/libarchive and dump,
so there is still stuff going on. We'd happily accept contributions to
improve deficiencies, ideally in the form of patches. :-)
Robert N M Watson
Computer Laboratory
University of Cambridge
More information about the posix1e
mailing list