Thinking more about the alias thing; an alias should be allowed to point
to a glue DSE. I couldn't find anything in X.501 that said an alias must
only point to DSETypes of entry or alias. Just that it points to an
object entry or an alias entry. It further confuses things when it says:
"The root is not an entry as such, but can, where convenient to do so
[e.g. in the definitions of b) and c) below], be viewed as a null object
entry [see d) below];" I assume it's not convenient to view the root as
a 'null object entry' when it's being referenced by an alias entry :-)
"Jim Sermersheim" <jimse@novell.com> 9/23/04 4:33:56 PM >>>
True, if there is a check to validate that aliasedObjectName points to
an entry or alias DSE, this should just result in an error.
So, getting back to my original question * does anyone know if X.500
allows access to the root DSE without the use of ManageDsaIT?
Jim
"Kurt D. Zeilenga" <Kurt@OpenLDAP.org> 9/23/04 3:33:34 PM >>>
At 02:16 PM 9/23/2004, Jim Sermersheim wrote:
I can reconcile some things, but not others. It's the inverse that is
the problem * in the absense of manageDsaIT, when is the root to be
treated as a normal object?
I *think* X.500 says never, and LDAP special cased read, and sort of
left the door open for special casing write.
I was hoping that an X.500 vendor with an LDAP front-end could clarify
(as they would have implemented the X.500 semantics, and the LDAP
front-end would have done something to special case where needed).
Another question is: Assume an alias object <dc=TopOfTheWorld> which
has an aliasedObjectName set to the empty DN.
One could consider this invalid. aliasObjectName is suppose to
refer an object entry (or possibly another alias entry). Having
an alias refer to a root DSE, which is not part of the DIT,
makes no (or, at most, little) sense.
If a base-scope search is rooted at that alias, the filter is
(objectclass=*), and derefAliases is derefFindingBaseObj, what is the
proper behavior? I think RFC 2251 could be read either way.
I would think the server would return some sort of aliasing
error here. (I believe the OpenLDAP server does.)
What I wanted to do with the distproc I-D was give a simple, straight
forward way of determining whether a DSE reresents a normal object (in
the absense of manageDsaIT).
Jim
"Kurt D. Zeilenga" <Kurt@OpenLDAP.org> 9/22/04 6:50:28 PM >>>
Though I said "I disagree" over on the LDAPBIS list, in
retrospect, I would have to minimally agree that a server
*could* returning the rootDSE in response to subtree
manageDsaIt search. RFC 3296 says:
The control MAY cause other objects to be treated as
normal entries as defined by subsequent documents.
While there is no subsequent document stating how this
control would cause rootDSE to be treated as a normal
entry, I think it reasonable that some subsequent document
could state how. I would hope that any such document
would state a how consistent with X.511(97) manageDsaIt
service.
Are you able to reconcile things if you assume that
the rootDSE is treated as a normal object in
manageDsaIT searches with baseObject=""?
Kurt
_______________________________________________
Ldapext mailing list
Ldapext@ietf.org
https://www1.ietf.org/mailman/listinfo/ldapext
_______________________________________________
Ldapext mailing list
Ldapext@ietf.org
https://www1.ietf.org/mailman/listinfo/ldapext
_______________________________________________
Ldapext mailing list
Ldapext@ietf.org
https://www1.ietf.org/mailman/listinfo/ldapext