On 22 Mar 2007, at 14:58, Josh Wisenbaker wrote:
On Mar 21, 2007, at 10:36 AM, Giuliano Gavazzi wrote:
Attempting the same for a user-based ACE did not appear to work
for a at first, but attempting the same after a few minutes at
least gives an appearance (illusion ?) of working. I'm not
saying this is a "good idea" by any means, as matching UUIDs
across disparate systems seems completely counter to the purpose
of a "Universally" unique identifier. Perhaps the an exception
could be an OD master and its slave.
except that according Apple this should work no matter what the
UUIDs are. I think they did not really understand the problem.
According to Apple each ACE is specific to a UUID. By design UUIDs
should always be unique to each object. They are generated by
combining the MAC address with the number of 100-nanosecond
intervals since October 15, 1582.
If you need consistent UUID resolution then you use a directory
service to provide that information.
except that this is not necessarily how rsync/scp and tar work. It
is conceivable that their behaviour, when users by the same
username exist on source and target, is meant to match users by
username and only fall to UUID when the match does not exist. This
is, mutatis mutandis (UUID <-> uid|gid), how rsync works by default.
My problem here is that many people talk about the(ir) theory but
do not seem to have done tests.
