[Samba] Rev #2 of the 3.02.3c patch

[Logan Shaw]

On Wed, 30 Aug 2006, Gerald (Jerry) Carter wrote:
> I've uploaded the *final* 3.0.23c roll up patch to
> http://samba.org/~jerry/patches/patch-3.0.23b-3.0.23c-gwc-2.diffs.gz.
> I've already cut the 3.0.23c tarballs so unless there is
> a major problem, this will be the final change set.
>
> Please report *any* bugs that you find.

Well, I'm not positive it's a bug, but with 3.0.23b, I can go
to the (Windows Explorer context menu) "Properties"->"Security"
dialog and look at the list of "Group or user names", and every
file I've tried shows the group's SID before the user's SID.

This isn't the order I expect, and in fact, it's different
from what I see on 3.0.10 system which runs against the same
ldapsam data.  With 3.0.10, I get what I expect:

 	Group or user names:
 	+-----------------------------------+
 	| (H)  Logan Shaw (MYDOMAIN\lshaw)  |
 	| (HH) engineer (MYDOMAIN\engineer) |
 	| (HH) Everyone                     |
 	+-----------------------------------+

(The "(H)" represents the single-human-head icon, meaning
user, and the "(HH)" represents the two-human-heads icon,
meaning group, I guess.)

With the 3.0.23b, I get something like this instead:

 	Group or user names:
 	+-----------------------------------+
 	| (HH) engineer (MYDOMAIN\engineer) |
 	| (HH) Everyone                     |
 	| (H)  Logan Shaw (MYDOMAIN\lshaw)  |
 	+-----------------------------------+

Note that the user appears at the bottom of the list.

I think this is probably related to something else I'm seeing:
when an Excel or Word file is open and locked by a user
and someone else tries to open it, they get a message that
it's locked by "engineer" (the group) rather than "lshaw"
(the username).

So, what relevance does this have to 3.0.23c?  Well, it's
happening with 3.0.23b, and I spent about 15 minutes looking
through the 3.0.23b->3.0.23c patch you just posted today
(patch-3.0.23b-3.0.23c-gwc-2.diffs.gz), and I couldn't see any
code changes that looked related.  Unfortunately, I can't really
take the server down to test the patched version for real.

Also, I'm fairly sure it didn't happen with 3.0.22 and that
I haven't changed smb.conf in any meaningful way since moving
from 3.0.22 to 3.0.23b, making me believe it's a function of
the samba version rather than the config.

So, I realize that's not the ideal bug report, but is it
possible someone running 3.0.23c could check and see if they
are seeing a similar issue?

   - Logan