RE: Listing request names

[Chris Weider <cweider@xxxxxxxxxxxxx>]

We should provide a unique-identification service for schema elements which
do not have them. But again, these can be irrelevant to specific protocols
which wish to use specific schema. And LDAP can use any naming tree
anywhere, and there's no imperative to put all schema under a single
subtree.
Chris

> -----Original Message-----
> From:	Michael Mealling [SMTP:michael@xxxxxxxxxxxxxxxx]
> Sent:	Monday, November 24, 1997 11:45 AM
> To:	Chris Weider
> Cc:	michaelm@xxxxxxxxxx; M.Wahl@xxxxxxxxxxxxxxxxxx; phoffman@xxxxxxx;
> ietf-schema-reg@xxxxxxx
> Subject:	Re: Listing request names
> 
> Chris Weider said this:
> > Object classes, templates, and attributes should have unique identifiers
> > assigned to them. Ideally these identifiers would not be human-friendly
> > names, because human-friendly names are quite likely to be reused. A
> given
> > protocol does not have to use the unique identifiers in its operations;
> for
> > example, whois++ can continue to use the human-readable attribute and
> > template names. But there needs to be some disambiguation possible other
> > than "well, it's the version of Name used by the whois++ server at
> > www.foo.com"....
> 
> I'm confused. I thought we were just listing schema, not providing
> a naming service. I was thinking of a system where each
> entity would end up having several different names: this is what LDAP
> calls this schema, this is what whois++ calls it, this is what the
> service itself calls it. But the listing service name has no more
> importance
> than any of the others. 
> 
> Anyway, I don't have a problem with the service creating OIDs. I just
> want them to be used by the service and not used as a directory services
> native naming scheme. I don't think we understand this stuff well enough
> to start figuring out how to name ALL schema. My personal opinion is that
> schema should be named with URNs and not OIDs. (yes, an OID would make a 
> valid URN and I even wrote a draft on it several years back).
> 
> I havn't gotten into LDAP schema stuff much so I'll ask a question out of
> shear ignorance: Is there a problem with LDAP schema using OIDs from 
> any subtree? Is there a desire in the LDAP community to start putting
> schema OIDs into a single standardized tree somewhere? 
> 
> > > > > Chris Weider said this:
> > > > > > Michael:
> > > > > >   The OID registration service is implicit in the listing being
> able
> > > > > > to assign names to the nameless.
> > > > 	[Chris Weider]  
> > > > > Sure. But are those the names that the schema is supposed to use
> > > within its own system? I.e. are the OIDs the service assigns going to
> the 
> > > the
> > > > > OIDs that LDAP uses everywhere?
> > > > 	[Chris Weider]  Yes
> > > > 
> > > > > By nameless do you mean schema's that have not yet been created or
> 
> > > > > schema that use names other than OIDs?
> > > >
> > > > 	[Chris Weider]  Yes, and for schema that do not yet have
> unique ids
> > > > assigned (e.g. Dublin Core).
> > > >
> > > 
> > > Yes to all of 'em? If whois++ wants to list its schema in the service
> it
> > > will have to rename its schema to the OID that the service assigns?
> > > 
> 
> 
> -- 
> --------------------------------------------------------------------------
> ----
> Michael Mealling	| 505 Huntmar Park Drive       | Phone:
> (703)742-0400
> Software Engineer	| Herndon, VA 22070	       | Fax:
> (703)742-9552
> Network Solutions	| <URL:http://www.netsol.com>  | michaelm@xxxxxxxxxx