W3C home > Mailing lists > Public > www-tag@w3.org > September 2003

RE: Liberal "URI assignment policies" will create confusion?

From: Williams, Stuart <skw@hp.com>
Date: Thu, 18 Sep 2003 12:30:51 +0100
Message-ID: <5E13A1874524D411A876006008CD059F04A077D2@0-mail-1.hpl.hp.com>
To: 'Ossi Nykänen' <onykane@butler.cc.tut.fi>
Cc: www-tag@w3.org

Hello Ossi,

> -----Original Message-----
> From: Ossi Nykänen [mailto:onykane@butler.cc.tut.fi] 
> Sent: 18 September 2003 10:34
> 
> My original "complaint" is that instead of (W3C) saying 
> something like "sure -- you MAY invent any URI spelling 
> (within your authority) as you wish", I would appreciate a 
> more constructive statement like "yes, we are actually 
> promoting consistent, long-term URI name bookkeeping (for 
> denoting certain kinds of Web resources); here are guidelines 
> how it could be done...". More formal the better.

I guess that the furthest I can go here is "'complaint' noted". 

AFAIK, we gave no existing, documented, widely applicable "long-term URI
name bookkeeping" to promote.

AFAIK, the W3C is not working on one.

The dominant disposition is that URI are opaque in general that within the
constraints of the delegation chain, URI assigment authorities are free to
any systematic means URI->resource mapping assignments.

Assignments authorities my choose adopt a particular style in assigning some
part their URI space. eg. they may declare that their URI conform to OpenURL
(which cropped up earlier on this topic); AFAIK, the W3C has no particular
position wrt to OpenURL (and that is not a W3C statement) nor is it
necessary that it have one IMO.

Have you looked at the approach WebDav have taken to versioning?

> Some resources are to be scrapped or versioned for sure -- to 
> me it would make lot of sense to (uniformly) recognise this 
> at the Webarch level (a cheap way might be an add-on to the 
> URI spelling; which of course is only one alternative).

The draft finding we're talking about is focussed solely on the encoding in
and extraction of metadata in URI. 

Except as licensed by specs. which could include specs. published by an
assignment authority wrt to URI assignments under their authority (ie
assignment policies)... the TAG takes the position that the observer of a
URI is not licensed to extract any metadata solely from the URI. There is
also a bias toward making use of as few specs./policies as possible - so
that in the limit URI are completely opaque (from the POV of an independent
observer).

> It 
> might be a bit tricky, of course, but then, few things 
> aren't. (Say, like profiling an ISO standard: now that's tricky.)
> 
> On Wed, 17 Sep 2003, Williams, Stuart wrote:
> 
> > ...
> > The content of an advertisment can change.
> > ...
> 
> True, but there are two cases: 1) changing an "existing" 
> advertisement A and 2) publishing a new one A' (and perhaps 
> throwing the old advertisement A away).
> 
> In other words, if the principle "Use URIs: All important 
> resources SHOULD be identified by a URI." is used, authors 
> are effectively forced to invent ad hoc versioning systems 
> (plural) in order to do URI bookkeeping (both A and A' to be 
> preserved). And the term "MAY" in 1.1.2 in [1] seems like 
> dodging the real issue.

The sense of this may is that authorities may publish their assignment
policies... ie. they may tell what they are (equally they may not). That
they have them is almost certain. That they make them visible is at their
descretion.

> Clearly both of the individual advertisements (A and A' in the case 2
> above) are important since they might be legally binding etc. 

Ok... so they assigned different URI... and indeed there may be a third
resource, say A'', which is the current season's advertisment - whereas A
and A' are a particular season's advert. This is much like the way the W3C
organises current version and specific version URI for its technical reports
see
http://www.w3.org/2003/05/27-pubrules.html#rules

> And in this case, the difference between a resource and its 
> representation is also pretty thin.

Well... some resources are static... once instantiated their state does not
develop over time - that would be true of a particular version of a
document. It would not be true of the current version of a document
(resource sense of the word document).

Also, there are potentially multiple representations available for the same
resource (for documents: HTML, DockBook, XMLSpec, XSL-FO, Latex, DVI,
Postscript, PDF, ... come to mind).

> And you are indeed right, sometimes I do get confused with 
> the terms like "thing", "object in the networked information 
> system", "resource", "representation", "metadata about the 
> representation", and "relationship between things". Well, 
> here are the definitions (of course, all of them are from WDs):
> 
> From [1] (Web-arch):
> 
> (a) Objects in the networked information system called 
> resources are identified by Uniform Resource Identifiers (URIs).
> 
> (b) A representation is data that represents or describes the 
> state of a resource. It consists of: [...] Electronic data 
> expressed in one or more formats [...and] Metadata about the 
> representation, such as the Internet Media Type [...]
> 
> From [2] (RDF Semantics):
> 
> (c) The [RDF] semantics treats all RDF names [i.e. URI 
> references or a typed literals] as expressions which denote. 
> The things denoted are called 'resources', following [RFC 
> 2396], but no assumptions are made here about the nature of 
> resources; 'resource' is treated here as synonymous with 
> 'entity', i.e. as a generic term for anything in the universe 
> of discourse.

In (c) I can only see the term 'entity' as potentially confusing since HTTP
is also uses the term 'entity' for something quite different.

> 
> And finally, from [3] (RDF Concepts):
> 
> (d) The assertion of an RDF triple says that some 
> relationship, indicated by the predicate, holds between the 
> things denoted by subject and object of the triple.

Ok... so here object maybe the problem, because the sense of the word here
(d) is different from the sense of the word in (a).

See http://www.cogsci.princeton.edu/cgi-bin/webwn1.7.1?stage=1&word=object

(a) uses sense 1 while (d) uses sense 3.

> 
> ...in other words you can't e.g. use RDF to describe the 
> representations (of resources) since by (d) RDF asserts 
> relationships between things, which by (c) are resources, and 
> by (a) and (b), resources and representations are different 
> (at least by your interpretation). 

Well that's a fine bit of navigation round the documents and completely
correct as far as I can see.

> Or then representations 
> indeed are (conceptually) resources, at least in certain 
> context ...to be considered as resources of certain type.

You may choose to create a resource whose sole purpose is to consistently
and persistently provide a single, static, immutable representation. In that
sense 'promote' a representation being a resource... although strictly the
resource and the represention remain different things.


> I don't want to see all the specs rewritten but yes: I am a 
> bit confused.

Actually... given the correct conclusion you came to above... I don't think
you're as confused as you (or I) might think.

> Cheers,
> 
> --Ossi
> 
> [1] http://www.w3.org/TR/webarch/ 
> [2] http://www.w3.org/TR/2003/WD-rdf-mt-20030905/
> [3] http://www.w3.org/TR/2003/WD-rdf-concepts-20030905/

regards

Stuart
--
Received on Thursday, 18 September 2003 07:32:10 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:20 GMT