Re: [ldapext] VLV: a particular target entry

["Jim Sermersheim" <jimse@novell.com>]

Given what David just said in his latest reply, if I were a client, I
would tend to assume that if the contextID is missing, the list _may_
become incosistent. But that neither insures that the list will become
inconsistent, nor that lists produced by servers returning a contextID
will remain consistent (in otherwords, implementations can be as clever
or sloppy as they want to be).

I think it's appropriate for the statement that discusses the intent of
the contextID to warn that servers not returning a contextID may produce
inconsistent results when paging. Is this what you're driving at? 

Jim.

>>> "Kurt D. Zeilenga" <Kurt@OpenLDAP.org> 10/03/02 03:10PM >>>
At 01:56 PM 2002-10-03, Jim Sermersheim wrote:
>The client may or may not run into a stability problem, depending on
the
>server implementation. I believe iPlanet implements VLVs by using a
>pre-canned result set (at least that's one implementation). With that
>model, it can better ensure consistency without using state
>information.

>I don't know how they handle changes to the canned result set, or
>changes to the users access control.

I don't see how they could possibly pre-can dynamic content.
I see how they might pre-can static content.

If no cookie is not returned, should the client assume the
content is static?

Kurt

_______________________________________________
Ldapext mailing list
Ldapext@ietf.org
https://www1.ietf.org/mailman/listinfo/ldapext