Re: Significant W3C Confusion over Namespace Meaning and Policy

On Feb 16, 2005, at 22:51, ext Harry Halpin wrote:

>
> While I agree that Henry is technically correct (technically as in  
> "read the specification"), this giant perma-thread clearly shows that  
> there are simply problems in keeping track of versioning with  
> namespaces. If a namespace can have an infinity of names, then perhaps  
> this should be repaired in a new spec, since having an infinity of  
> names makes versioning
> difficult.


Only if you are focusing on the wrong thing at the wrong level.

If you presume that there is a 1:1 relationship between
namespace and vocabulary and/or schema, then yes, you will
have no end of headaches when new terms are added to a
given namespace.

But if you accept that vocabularies, schemas, ontologies,
and other similar models may employ terms from multiple
namespaces, and share terms between different versions
and models, and that core application logic should not be concerned
with syntactic issues such as namespaces, then versioning becomes
far more straightforward and managable.

Focus on the models, and leave the namespaces in the syntax.

> Either one has a separate URI for each version, or a representation is  
> returned by the URI that contains the versioning information. For  
> example, the RDDL representation retrieved by  
> http://www.w3.org/XML/1998/namespace
> clearly demarcates what version of XML is being used, as well as  
> elements
> (although no xml:id as of yet).
> Since a RDDL is both machine and human-readable (and there's even RDDL  
> RDF), this would be a good mechanism
> for keeping track of the actual finite number of name authorized by the
> resource owner for use in the namespace. However, while RDDL is nice,
> the W3C has yet to standardize it or some variant for this purpose.

I'm all for both machine and human readable information, and easy
and straightforward discovery of information. But RDDL is not the  
answer.

Rather RDDL, or rather, the idea of namespace documents, should just be  
dropped.
Getting back a namespace document which identifies all of the versions  
of all of
the vocabularies, schemas, ontologies, etc. employing terms from that  
namespace
is useless.

How is an application to know *which* version of *which* model to
apply in order to interpret the term in question? It can't, because
a given namespace is not tied to a specific version of a specific
model -- insofar as the specifications are concerned (and that's
the key point!). Yes, *some* version of *some* model may utilize
a *single* namespace for *all* terms employed -- but that is a
proprietary, localized practice and not a practice that (a) any
agent can presume will be guarunteed and (b) could never be mandated
on a global scale (at this stage in the game).

Thus, the only way that namespace documents could offer any reliably
utility to arbitrary web agents for arbitrary web content would be
for the W3C to mandate that all namespace names always identify
namespace documents -- and since that will never happen, the approach
offers no solution (and as an aside, discussion of namespace documents
should never have been included in AWWW and should be removed in
future editions).

Far better to explicitly identify each version of each vocabulary,  
schema,
ontology, etc. with a URI and indicate in the instance of the data which
version of which model it conforms to.

(as to merely well-formed but otherwise unqualified XML, well, agents
will just have to guess anyway...)

>
> This is awkward because in the AWWW it is noted that:
>
> http://www.w3.org/TR/2004/REC-webarch-20041215/#representation- 
> management
>        "A URI owner may supply zero or more authoritative  
> representations
>  of the resource identified by that URI. There is a benefit to the
>  community in providing representations.
>
>        Good practice: Available representation
> A URI owner SHOULD provide representations of the resource it  
> identifies"
>
> So, in summary, a URI owner may serve zero representations, but they  
> should  server at least one. Personally, I think emphasis on "should"  
> needs to be increased,

No, because increasing it would make it a MUST, and the W3C cannot
(successfully) make such a strong mandate about URI usage.

There are numerous valid uses of URIs which do not require providing
representations of the resources identified by those URIs (however
useful it might be) and since at the end of the day, someone has to
*pay* to create and manage representations, if there is no motivating
need to do so, it should not be considered "wrong" not to.

AWWW is perfectly correct in its use of SHOULD.

> and one way would to propose a standard
> for versioning namespaces using RDDL or something like it. I don't
> believe this has been done, and I'm not sure why - since it would be a
> great boon to the community.

Rather, let those producing content indicate the models governing the
interpretation of that content, using URIs identifying those models,
and via those URIs, agents can then obtain representations of those
*models* and thus the information needed to properly interpret the
content.

Cheers,

Patrick



>
> 				Cheers,
> 					harry
>
>
>  On Wed, 16 Feb 2005, Bullard, Claude L (Len) wrote:
>
>>
>> End of story?  Sure.  End of problem?  No.
>> The versioning saga goes on.
>>
>> len
>>
>>
>> From: www-tag-request@w3.org [mailto:www-tag-request@w3.org]On Behalf  
>> Of
>> ht@inf.ed.ac.uk
>>
>> It follows that _as defined in the Namespace REC_ there are infinitely
>> many names in every namespace.
>>
>> End of story.
>>
>>
>
>
> 	Harry Halpin
> 	Informatics, University of Edinburgh
>         http://www.ibiblio.org/hhalpin
>

Received on Thursday, 17 February 2005 08:40:00 UTC