W3C home > Mailing lists > Public > www-rdf-interest@w3.org > November 2003

Re: RDF query and Rules - my two cents

From: NMP-MSW/Tampere <patrick.stickler@nokia.com>
Date: Wed, 19 Nov 2003 10:28:47 +0200
Cc: www-rdf-rules@w3.org, www-rdf-interest@w3.org
To: "ext Mark Baker" <distobj@acm.org>
Message-Id: <64F9A160-1A6A-11D8-B3FE-000A95EAFCEA@nokia.com>


On Tuesday, Nov 18, 2003, at 19:33 Europe/Helsinki, ext Mark Baker 
wrote:

> On Tue, Nov 18, 2003 at 12:09:36PM +0200, Patrick Stickler wrote:
>>> I think a suitable approach would be to build on the existing RDF
>>> remote
>>> access API - that of RDF/XML+HTTP. A http GET will retrieve a model
>>> over the
>>> network based on a supplied URI. The RESTful continuation would start
>>> with a
>>> PUT to place it on the network, DELETE remove it.
>>
>>
>> I consider this far too coarse grained to be efficient and generally
>> useful
>> (note the important qualification 'generally').
>
> Well, there's a difference between the logical operation and what goes
> out "on the wire".  One could use RFC 3229 to send & request deltas, 
> for
> example.
>

That presumes that the client wants, or is even capable, of recieving
the entire model. If all a mobile client needs is knowledge about a
particular resource described by a model which in total is 500MB in
size, yet the description of the resource of interest is only 1K in
size -- and the mobile device wishes to (a) obtain the remotely managed
description, (b) modify it locally, and (c) update the remote KB,
it should be able to do so without having to GET and PUT the whole
model.

I could also argue for scalability and network load efficiency even
for desktop clients.

The bottom line is this: any standardized query protocol should allow
for the interchange of arbitrary subsets of any kb irrespective of
the internal organization of that kb -- whether it be a single physical
database, a single RDF/XML instance, a virtual cluster of 
databases/servers,
a filesystem of RDF/XML instance files, etc. etc.

RFC 3229 doesn't solve this problem, since it requires having a local
copy of the *whole* representation, which (even for many desktop 
clients)
is not IMO reasonable or useful.

Yes, some folks may want/need to move entire models around, but they
are, I think, the exception rather than the rule insofar as general
RDF interchange is concerned. And yes, for them, HTTP GET and PUT should
work just fine, so there's nothing to be done there, eh?

Regards,

Patrick
Received on Wednesday, 19 November 2003 03:31:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:03 GMT