svn commit: samba r24465 - in branches/SAMBA_3_2_0/source/libsmb: .

[Gerald (Jerry) Carter]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Zack,

> I have to disagree here.  Coming from a file system 
> that has builtin NTFS ACL support, we have the same
> problem of whether or not to sort/canonicalize the ACL.
> There is no clear "right" solution, as both
> sorting and not sorting have advantages and disadvantages.
> 
> Obviously, sorting the ACL makes viewing them easier 
> on Explorer. However, sorting the ACL has some
> disadvantages; Samba will not be relaying the exact permissions
> to Windows (i.e. lying) when a Posix ACL  or Posix mode is
> on the file.  An even bigger problem with sorting the
> ACL is if a user modifies the ACL (i.e. adds an ACE) and 
> then the sorted ACL becomes the real ACL.  Finally, it is
> possible that a Windows client has explicity set a
> non-canonicalized (non-sorted) ACL itself -- shouldn't
> Samba be able respect this, even if Explorer complains?

Sorry, but aren't we talking about libsmbclient here.  What
Gnome developers do you know that care about ACL sorting?   If
you want the most flexibility, then add a function ptr to
your own sorting function.  But by default an API should be
hard to misuse.

If I understand the bug report, there was a problem setting ACLs
via libsmbclient due to misordering.  Derrell's change fixes that.

So before I keep stepping up to bat on this one, someone
please confirm that I followed the original thread correctly.
Thanks.



cheers, jerry

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGw4fZIR7qMdg1EfYRAuDHAKCj3HACOGPIwsipQLsNB2+h4DUBmQCeJPAT
3tX9CEAbq9267+0u2Q066NI=
=EJ8O
-----END PGP SIGNATURE-----