RE: About aliases an X.500....

[Alan Lloyd <Alan.Lloyd@OpenDirectory.com.au>]

> -----Original Message-----
> From:	Ed Reed [SMTP:ED_REED@novell.com]
> Sent:	Wednesday, April 08, 1998 2:06 AM
> To:	Harald.Alvestrand@maxware.no; howes@netscape.com;
> ietf-ldapext@netscape.com; Alan.Lloyd@OpenDirectory.com.au
> Subject:	Re: About aliases an X.500....
> 
> NDS implements aliases in such a way that the naming rule for the
> alias is _identical_ to the naming rule that applies to the referenced
> object.  For instance, it seems meaningless to us to allow an alias
> which points to a Person thing to use anything other than the same
> naming attributes as that person.  Ditto containment (can't put a
> alias to a Person in places you couldn't put the person in the first
> place).  Similarly, of course, you can't "cast" a Person into a Server
> by changing the ObjectClasses list for the alias...the alias reflects
> exactly what is defined on the instance of the object to which it
> points.
> 
> I agree that aliases can be useful in constructing alternative
> namespace hierarchies for browsing or even searching.  But it makes my
> head hurt thinking too much about changing the object class types,
> naming rules, or containment rules which govern aliases to be
> different from the objects to which they refer.
> 
> Ed
	Totally agree Ed. Directories are about disciplined information
systems - which are controlled by the structure rules. X.500 lets alias
to be used according to structure rules as configured.
	Beer also causes heads to hurt! :-)
	regards
> -------------------
> Ed Reed, Technologist
> Group Technology Office
> Novell, Inc.
> +1 801 861 3320
> 
> >>> Harald Tveit Alvestrand <Harald.Alvestrand@maxware.no> 04/07/1998
> 01:13:39 >>>
> Alan,
> on a completely different topic:
> I've thought for a long time that using aliases is the only way
> one can sensibly use the distinguished name for uniqueness and for
> search scope limitation at the same time.
> 
> I think I have also got a fairly solid gut feeling for how it's being
> used
> (basically adding empty records of type "alias", with only the naming
> attribute; assuming that searches follow aliases encountered in a
> subtree traversal unless the appropriate control is invoked).
> 
> But I was trying to find supporting text for this picture in the X.500
> series when I came across this entry in X.501(93) 12.3.3:
> 
> >NOTE - The object class alias does not specify appropriate attribute
> types
> >for the RDN of an alias entry. Administrative Authorities may specify
> 
> >subclasses of the class alias which specify useful attribute types
> for RDNs
> >of alias entries.
> 
> Put this together with the fact that X.520 does not mention aliasing,
> and that naming rules don't seem to give either permission or
> restriction
> on the use of RDNs for aliases, and it seems to me that the standard
> and
> its usage is somewhat underspecified.
> 
> I see two possibilities:
> 
> - The industry has ignored naming rules where aliases are concerned,
> and
>   is buliding products that "allow anything".
> - The industry is shipping private schemas that specify rules for how
> one
>   can use aliases, but these are not standardized.
> 
> Obviously, software developed to the two paradigms above will not
> necessarily
> interoperate.
> 
> Can you help me understand this one?
> 
> Regards,
> 
>                          Harald T. Alvestrand
> 
> -- 
> Harald Tveit Alvestrand, Maxware, Norway
> Harald.Alvestrand@maxware.no 
>