W3C home > Mailing lists > Public > public-awwsw@w3.org > June 2009

Re: Are generic resources intentional?

From: Jonathan Rees <jar@creativecommons.org>
Date: Tue, 9 Jun 2009 08:40:05 -0400
Message-ID: <760bcb2a0906090540o59938db7g9b9f92646f5e3864@mail.gmail.com>
To: noah_mendelsohn@us.ibm.com
Cc: AWWSW TF <public-awwsw@w3.org>, "Williams, Stuart (HP Labs, Bristol)" <skw@hp.com>
The question is answered by RF 2616 7.1 "Entity Header Fields", and
the answer is yes (i.e. Content-location is an entity-header field,
thus part of the wa-representation).

I think you're being a bit too concrete. I think the situation Tim's
trying to enable is distinct generic-resources with the same trace -
these would be named with the same URI, but I think you could get
identical traces for distinct URIs, even if Content-location is
specified. Remember content-location is not just another name for the
same resource. The two GRs could reside on the same server, and that
server could "intern" its representations, noticing that the reps of
these two GRs are coincidentally the same, and return the same
Content-location for the shared representation. As far as I can tell
this would be completely correct.

I think you might be arguing for the situation I would prefer in an
ontology, where the only differences in web resources are those that
are in some sense observable (that would include Content-location), as
opposed to Tim's view, which is that generic-resources can differ in
spite of there being no observable difference. To me this contradicts
the "essentially information" and "conveyable in a message" ideas. The
meaning of a book is a process in which the author and reader
participate; it is not part of the identity of the book, which is only
an instrument in that process.

(Personally at this point I think that regarding web architecture or
HTTP semantics I would ditch all the philosophy about "essentially
information" and "conveyable in a message" and just stick to something
much more operational and concrete. I'm not sure what that would be;
maybe start with "on the web" or "can be put on the web" or "suitable
for use with HTTP" and try to tighten that up, or else just admit,
along the lines suggested by Harry, that the ontology aspect is both
hopeless and silly, or David, who says it's out of scope, and ask
people to do the best they can with the use cases we already have. (By
"ontology" here I mean a theory of what things exist and what they're
like, not a document type.)  Then separately encourage principled
ontologies, such as FRBR or IAO, not directly related to HTTP or the
Web, to account for  domains they care about.)

Jonathan

On Fri, Jun 5, 2009 at 10:02 AM, <noah_mendelsohn@us.ibm.com> wrote:
> Just curious:  If Content-location is returned, does that count as part of
> the trace?  If so, then it seems that this issue only comes up in cases
> where the Content-location is not supplied.
>
> Noah
Received on Tuesday, 9 June 2009 12:48:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 9 June 2009 12:48:56 GMT