Re: [ebxml-cppa] CPP identification revisited

[Christopher Ferris <chris.ferris@sun.com>]

URN might be an acceptable approach. However, I think it
needs some further thought.

For instance, a URN such as
	urn:oasis-open.org:tc:regrep:<UUID>
might on the surface seem to provide acceptable context, but
given that the reg/rep isn't universally replicated, the
scheme above doesn't provide the requisite context regarding
*which* reg/rep assigned the UUID and hence, no ability to
cross reference artifacts that were registered in a foriegn
registry from the artifact which references it.

Imposing a requirement that each instance of a reg/rep
acquire its own URN namespace seems excessive. Use of
the HTTP URI scheme combined with the domain name of
the assigning authority (qualified by an optional
path e.g. http://<mydomain>/regrep/resolver/<UUID>) would
be the most web friendly approach to the problem at
hand, impose the least amount of administrative overhead
and at the same time provide sufficient context to
enable the resolution of the artifact which is identified.

Again, I think it important that we provide for private
registries which may have need to reference registered artifacts
in a public or foriegn registry.

The only other approach I might offer would be that any artifact
which references a foriegn-assigned UUID MUST provide the
capacity to resolve that UUID to the artifact registered
elsewhere, by whatever means necessary. This could be considered
the equivalent to a symlink in a filesystem namespace that
references an NFS-mounted filesystem.

Cheers,

Chris

Duane Nickull wrote:

> 
> "Sharma, Nita" wrote:
> 
>>1. Meaningful verses meaningless identification scheme
>>
> 
> Nita:
> 
> "Meaningful" is not a requirement for the UID.  The UID is to be
> interpretted by applications.  We have other  ways to convey
> semantically meaningful element identifiers such as the Element names in
> XML.  The only requirement for a UUID is that is is globally, uniquely
> identifiable.   
> 
> 
>>2. multiple identification scheme for the same item based on various usages.
>>
> 
> The real issue is not as much this as the fact that we need three things
> to retrieve a Managed Object during the Discovery Phase. 
> 
> 1. A UUID
> 2. the location where you can retrieve it
> 3. the protocol for the retrieval binding (eg HTTP, ebXML RSS + ebXML
> MHS).
> 
> 
>>3. A standard organization (like UCC/EAN) to control the uniqueness and
>>meaningfulness
>>
> If by "meaningfulness" you mean that it has semantic meaning to you (or
> any other human being),  then I urge you to revisit.  The UUID has one
> basic task.
> 
> We may wish to use the URN format.  Dale M. forwarded several excellent
> RFC's written by Michael Mealling and others to the CPP/a list. 
> 
> An in term solution would be to use URN's like Commerce One did with a
> proprietary resolution mechanism using DNS.  IN other words a solution
> similar to  a backwards domain name of the regsitry, followed by a colon
> then the actual UUID the registry assigned.  This is not a good solution
> but it could be usable in the mean time.
> 
> Duane
> 
> 
>>4. Separate namespace for the various identification schemes
>>5. Analyze other schemes like IDEF, OID etc.
>>
>>We in the catalog team are working out the use cases for identifications
>>
>>Cheers
>>
>>- Nita
>>
>>-----Original Message-----
>>From: Nikola Stojanovic [mailto:Nikola.Stojanovic@encodasystems.com]
>>Sent: Monday, February 11, 2002 1:40 PM
>>To: 'Duane Nickull'; Martin W Sachs
>>Cc: ebxml-cppa@lists.oasis-open.org; 'regrep@lists.oasis-open.org'
>>Subject: RE: [ebxml-cppa] CPP identification revisited
>>
>><Duane>
>>The Registry assigned unique identifier identifies it uniquely within one
>>specific registry instance, not globally.
>></Duane>
>>
>>>From the OASIS/ebXML Registry Services Specification v2.0
>>http://www.oasis-open.org/committees/regrep/documents/2.0/specs/ebrs.pdf
>>lines 648-650
>>
>><snip>
>>As specified by [ebRIM], all objects in the registry have a unique id. The
>>id must be a Universally Unique Identifier (UUID) and must conform to the to
>>the format of a URN that specifies a DCE 128 bit UUID as specified in
>>[UUID].
>></snip>
>>
>>Being UUID it is Universally (Globally) Unique. Some call it GUID (Globally
>>Unique).
>>
>>Regards,
>>
>>Nikola Stojanovic
>>Lead Technologist, Research and Development
>>Encoda Systems, Inc.
>>101 Pineview Terrace
>>Ithaca, NY 14850 USA
>>nikola.stojanovic@encodasystems.com
>>Tel: 607-273-2224
>>
>>----------------------------------------------------------------
>>To subscribe or unsubscribe from this elist use the subscription
>>manager: <http://lists.oasis-open.org/ob/adm.pl>
>>
>