Re: Generic-Resources-53: URIs for representations

On Oct 4, 2006, at 14:03, ext Booth, David (HP Software - Boston) wrote:

> I agree there is no problem from a theoretical perspective, but it  
> would
> be awful from an expository perspective.

I think that hinges on whether one is trying to make the web machinery
"do semantics" or not.

If one takes the view, as I do, that the web layer is only concerned
with representations of resources, and the actual nature and identity
of those resources is irrelevant insofar as providing reliable and
consistent access to representations via URIs, then it is sufficient
to simply define the class of atomic Representations and note that

(a) at the web level, one can never know if a given URI denotes a
     representation or not, even if it invariably always resolves
     to the same byte sequence

(b) if is known (by extra-web means) that a given URI denotes a
     representation, then the agent is licensed to expect that every
     time it dereferences that URI it will get the same exact byte
     sequence

(c) determining whether a given URI denotes a particular class of
     resource, representation or otherwise, is the domain of the
     semantic web (or human/social interaction)

and finally, and perhaps most importantly,

(d) the web machinery need not care what a given URI actually denotes,
     only representation(s) of whatever resource might be denoted by  
that
     URI are made accessible via that URI

Yes, there are many social and process issues that depend on software
agents and/or humans knowing and agree on what resources particular URIs
denote, and how those resources relate to other resources, but all of
that is outside the scope of the web machinery proper -- and much of  
that
is specifically the scope of the semantic web.

In short, whether a URI denotes a representation or some other class of
resource is irrelevant at the web layer, and easily addressed at the
semantic web layer, so unless you try to push the semantics of URIs down
into the web layer, there is no problem.

Cheers,

Patrick


>
> David Booth, Ph.D.
> HP Software
> dbooth@hp.com
> Phone: +1 617 629 8881
>
>
>> -----Original Message-----
>> From: Patrick Stickler [mailto:patrick.stickler@nokia.com]
>> Sent: Wednesday, October 04, 2006 2:58 PM
>> To: Booth, David (HP Software - Boston)
>> Cc: Dan Connolly; www-tag@w3.org; raman@google.com; Williams,
>> Stuart (HP Labs, Bristol)
>> Subject: Re: Generic-Resources-53: URIs for representations
>>
>>
>> If one considers the representation as the atomic unit of the web,
>> then one can say that a representation is a discrete stream of
>> bytes and a representation of a represenation is always bit-equal
>> to itself.
>>
>> If one has a URI which the URI owner asserts denotes a  
>> representation,
>> then that's what it denotes, and dereferencing it should result in
>> getting back that discrete sequence of bytes that is the
>> representation
>> in question. A URI which supposedly denotes a representation would
>> always
>> return the exact same sequence of bytes every time it is  
>> dereferenced.
>>
>> So no problem with a representation being denoted by a URI, and one
>> need not worry about having "turtles all the way down".
>>
>> C.f. http://lists.w3.org/Archives/Public/www-rdf-interest/2004Oct/
>> 0076.html
>> for more thoughts on this.
>>
>> Cheers,
>>
>> Patrick
>>
>>
>>
>> On Oct 4, 2006, at 13:40, ext Booth, David (HP Software -
>> Boston) wrote:
>>
>>>
>>>> From: www-tag-request@w3.org [mailto:www-tag-request@w3.org]
>>>>
>>>> On Oct 2, 2006, at 7:31 AM, Williams, Stuart (HP Labs, Bristol)
>>>> wrote:
>>>>> I think that in creating webarch [2] we tried to maintain a fairly
>>>>> clear distinction between resources and representations (modulo
>>>>> anything can be a resource!). In that world view, IIRC, it
>>>>> was "resources" rather than "representations" that have URIs.
>>>>
>>>> Er... you just noted yourself that anything can be a resource.
>>>>
>>>> As such, I think it's not too harmful to speak of URIs for
>>>> representations.
>>>
>>> I rather strongly disagree.  For one thing, it is apt to lead to
>>> unnecessary confusion.  For another thing, once one of those
>>> representations has a URI and it responds to http requests,
>> we'd be in
>>> the rather uncomfortable position of having to refer to what it
>>> returns
>>> as "the representation of the representation" -- not a direction I
>>> think
>>> we want to go.
>>>
>>> Please find some other term to use.  I'll just through out
>> a few more
>>> ideas to stimulate thinking:
>>>
>>> 	snapshot
>>> 	view
>>> 	custom view
>>> 	presentation
>>>
>>>>
>>>>> [1]
>>>> http://www.w3.org/2001/tag/doc/alternatives-discovery-20060915.html
>>>>> [2] http://www.w3.org/TR/webarch
>>>
>>>
>>> David Booth, Ph.D.
>>> HP Software
>>> dbooth@hp.com
>>> Phone: +1 617 629 8881
>>>
>>>
>>
>>

Received on Wednesday, 4 October 2006 19:29:19 UTC