RE: caching HTTP 303 responses

Hello Jacek,

> -----Original Message-----
> From: Jacek Kopecky [mailto:jacek.kopecky@deri.org] 
> Sent: 12 July 2007 12:40
> To: Williams, Stuart (HP Labs, Bristol)
> Cc: Jeremy Carroll; Giovanni Tummarello; semantic-web@w3.org
> Subject: RE: caching HTTP 303 responses
> 
> Hi Stuart, I'm afraid there might be some misunderstanding between us,

Actually I don't think so...

> On Thu, 2007-07-12 at 12:24 +0100, Williams, Stuart (HP Labs, Bristol)
> wrote:
> > > For instance the DC case - they have different URIs, and initially

> > > the agent doesn't know anything about them. It dereferences 
> > > dc:description, let's say, and finds some information about that
and 
> > > other DC properties. It can probably assume now that it  need
never 
> > > again (for small values of never) dereference that. But when it 
> > > encounters dc:title, how can the client know that the stuff it got

> > > from the dc:description redirect is all the pertinent information 
> > > that it can get from dc:title, which it doesn't yet know to
redirect 
> > > anywhere?
> > 
> > Well... does it know *enough* about dc:title not to have to ask the 
> > question? In an open-world semantic web, finding *all* of anything
is 
> > presumably hard - and determining pertainancy. Maybe, one should
take 
> > the view that if your going to say something about something you 
> > should say all that you consider pertainent about that thing in the
one breath.
> 
> The client has never dereferenced dc:title, and that's its 
> primary way of getting to know about it unless it has some 
> kind of "authoritative ontology for the DC namespace". It has 
> dereferenced dc:description and learned something about 
> dc:title, but it has not learned that it has learned enough.

What makes something authoritative is an interesting question...

What client a consequence of asking about dc:description is a pile of
triples that, in addition to having something to say about
dc:description also have something to say about dc:title. Those triples
have an origin (http://dublincore.org/2006/12/18/dces.rdf) which it may
or may not choose to trust. If it chooses to believe the origin over
additional items and the origins  said all that it deems to be
pertainent about those things... then you win.

That aside, Roy's response [1] on www-tag, I guess certainly offers some
relief how HTTP should work (assuming that there is concensus that he is
right).

> Perhaps an assertion like this:
> 
> <owl:Ontology about="">
>   <ex:authoritativeForNamespace rdf:resource="&dc;"/> </owl:Ontology>
> 
> The meaning of namespace here is "all URIs that begin with 
> the given URI", I guess, or something similar, based on the 
> syntax of the URIs.
> It feels a bit dirty to dig into the URIs, but we should take 
> advantage of the common case that URI hierarchy actually 
> makes sense. 8-)

I wasn't thinking of such URI hackery, more just choosing to believe (in
good faith) what you've heard in response to another question from a
source that you trust at least in respect of the question that you
asked.

> 
> > FWIW, I sympathise and not being able to use a local cache to avoid 
> > the round trip it's hard. I think that you can certainly avoid the 
> > round trip wrt to repeated reference to the *same* thing.
> 
> Certainly, I wasn't even considering trying to dereference 
> dc:description twice here.
> 
> Jacek

Stuart
--

Received on Thursday, 12 July 2007 13:02:00 UTC