Re: new package: devel/acl

[Greg Troxel <gdt%lexort.com@localhost>]

David Holland <dholland-pkgtech%netbsd.org@localhost> writes:

> On Sat, Apr 08, 2023 at 11:25:36PM +0200, Alexander Schreiber wrote:
>  > The first is devel/acl - tools to manipulate POSIX ACLs as well as
>  > library/headers to write other programs to do so. Hard build dependency
>  > for filesystems/glusterfs on Linux.
>  > 
>  > Package was built & tested on Linux amd64, included binaries work,
>  > and glusterfs builds against. This will _not_ build on NetBSD due to
>  > too many "All the world is Linux" assumptions (and has been marked
>  > so in the Makefile).
>
> Looking at the code, this appears to be a library for implementing
> POSIX ACLs using extended attributes, so linux-specific or not it
> isn't applicable to NetBSD where the system already has POSIX ACLs.
>
> Is that right? Should I go update the DESCR and COMMENT?

My impression is that there are "POSIX ACLs" which are compliant with a
withdrawn draft, and there are "NFSv4 ACLs" which have a spec and are a
superset.   And then that Linux implements the semantics of POSIX ACLs
(fine), stores them in os-specific EAs (ok) but either does not
implement the withdrawn-standard interface or nobody on linux uses it
(not ok).

I think this might be a package that provides the POSIX interfaces, so
that programs written to POSIX will work on Linux.  That implies that
GNU/Linux systems tend to have a non-standard ACL interface,

If that's true, then this is

  Implement POSIX ACLs using Linux's not-really-POSIX ACL implementation

and it can be ONLY_FOR.

If it is trying to do something that isn't "implement the standard
interface", then it's a good question why it isn't BROKEN_ON instead.


Howver, all my attempts to understand this so far have felt like "You
are standing at a crevice.  You are holding a document that specifies
POSIX ACLs, which weighs several pounds.  There is a POSIX language
lawyer standing in a shadow nearby, looking menacing, and to the north
is a maze of twisty passages, all alike."