3.0.21pre2/rc1 and acl compatibility, deleted function in parse_prs.c

[Thomas Bork]

Jeremy Allison wrote:

>>acl compatibility (S)
>>This parameter specifies what OS ACL semantics should be compatible 
>>with. Possible values are winnt for Windows NT 4, win2k for Windows 2000 
>>and above and auto. If you specify auto, the value for this parameter 
>>will be based upon the version of the client. There should be no reason 
>>to change this parameter from the default.
>>Default: acl compatibility = Auto
>>Example: acl compatibility = win2k
>>mg2 1.2.0cvs2005-11-10 # testparm -sv 2>/dev/null | grep 'acl comp'
>>        acl compatibility =
>>mg2 1.2.0cvs2005-11-10 # grep 'acl comp' /etc/smb.conf
>>mg2 1.2.0cvs2005-11-10 # grep -r 'acl compatibility' 
>>/usr/src/SAMBA_3_0_RELEASE/source
>>/usr/src/SAMBA_3_0_RELEASE/source/param/loadparm.c:     {"acl 
>>compatibility", P_STRING, P_GLOBAL, &Globals.szAclCompat, 
>>handle_acl_compatibility,  NULL, FLAG_ADVANCED | FLAG_SHARE | FLAG_GLOBAL},
>>/usr/src/SAMBA_3_0_RELEASE/source/param/.svn/text-base/loadparm.c.svn-base: 
>>    {"acl compatibility", P_STRING, P_GLOBAL, &Globals.szAclCompat, 
>>handle_acl_compatibility,  NULL, FLAG_ADVANCED | FLAG_SHARE | FLAG_GLOBAL},
>>Seems the default is *not* auto.
>>If setting 'acl compatibility = auto' in smb.conf:
>>mg2 1.2.0cvs2005-11-10 # grep 'acl compatibility' /etc/smb.conf
>>acl compatibility = auto
>>mg2 1.2.0cvs2005-11-10 # testparm -sv 2>/dev/null | grep 'acl compatibility'
>>        acl compatibility =
>>mg2 1.2.0cvs2005-11-10 #
> Internally the string "" is treated as "auto". It's confusing I know
> but the internal code is doing the right thing even if it's not printed
> that way in testparms.

It's really confusing :)
Would it be better to change testparm to print out the real value 'auto' 
  if the setting is 'auto'?
The unwritten law says, that all parameters are explicit empty (not 
set), if testparm is showing their ending with '...='.
At the moment the behavior of 'acl compatibility' is so, that testparm 
shows it as empty, if it is really set to 'auto'.

>>Second a question @Jeremy:
>>You added the function only back to trunk and not to SAMBA_3_0:
>>http://websvn.samba.org/cgi-bin/viewcvs.cgi?rev=11486&view=rev
>>http://websvn.samba.org/cgi-bin/viewcvs.cgi?rev=11446&view=rev
>>Thats why the function is also not in SAMBA_3_0_RELEASE for the upcoming 
>>3.0.21rc1:
> Not needed in SAMBA_3_0, only HEAD at the moment.

Thanks for the clarification.

-- 
der tom
[fli4l-/eis-team]