RE: Alias pointing to ancestor

["Ramsay, Ron" <Ron.Ramsay@ca.com>]

Thomas,

The centre of your argument here is how search results are collated in the
presence of aliases. In this context it probably doesn't matter if they
point up or point down.

I think Jim was referring to the navigation issues of a chain of aliases
actually pointing to a member of the chain. This was addressed in X.500 with
the error aliasProblem.

Ron.

-----Original Message-----
From: Salter, Thomas A [mailto:Thomas.Salter@unisys.com]
Sent: Thursday, 25 January 2001 7:07
To: Jim Sermersheim; ietf-ldapext@netscape.com
Cc: d.w.chadwick@iti.salford.ac.uk
Subject: RE: Alias pointing to ancestor


(I glad to see you copied David Chadwick.)  In his book, Understanding
X.500, he states that the '93 standards allow for circular references and
require the DSA to avoid looping by keeping track of which parts of the tree
have already been searched.  (Section 5.10, p.149)

I can't find an explicit reference in the X.500 standards.  The closest
thing I found is the flowchart in Figure B.10 in Annex B of X.511.  Next to
a decision box labeled "alias object name to be considered?" is the note
"e.g., not already included; does not lead to search loop".

It thus appears the intent was to allow circular references.


A related question that came up in interoperability testing some time ago
had to do with order of the evaluation of search filters and dereferencing
of aliases.  The flowchart B.9 in X.511 clearly indicates that aliases are
dereferenced before the search filters are applied.


 > -----Original Message-----
 > From: Jim Sermersheim [mailto:JIMSE@novell.com]
 > Sent: Wednesday, January 24, 2001 2:17 PM
 > To: ietf-ldapext@netscape.com
 > Cc: d.w.chadwick@iti.salford.ac.uk
 > Subject: Alias pointing to ancestor
 > 
 > 
 > A question has risen in the context of some interoperability 
 > tests regarding whether it is allowed to have an alias entry 
 > that points to an ancestor entry.
 > 
 > I can't find any reference to this in X.501. I see in 
 > section 9.5 that an alias can point to a non-leaf, as well 
 > as to another alias, but it doesn't mention pointing to accentors.
 > 
 > I assume it's disallowed, otherwise there should be an 
 > instruction regarding how circular dereferencing is to be 
 > handled. I couldn't find anything like that in X.511.
 > 
 > Is any of this talked about and I'm just missing it?
 > 
 > Jim
 >