Re: re-opening the DESCRIBE issue

On Thu, Mar 30, 2006 at 04:29:19PM +0100, Seaborne, Andy wrote:
> 
> 
> 
> Dan Connolly wrote:
> >In reviewing our request for CR,
> >  http://www.w3.org/2001/sw/DataAccess/crq349
> >there was a lot of concern about interoperability
> >around DESCRIBE. TimBL took a close look at CBD
> >and MSG from
> >http://semanticweb.deit.univpm.it/submissions/www2005/WWW2005_signignRDF.pdf
> >and looked briefly at GK's recent proposal
> > http://www.w3.org/mid/4427D376.9070504@ninebynine.org
> >and concluded that this issue merits considerable further work.
> >We talked about the possibility of postponing the issue
> >and marking the DESCRIBE syntax "reserved for future use"
> >and he was supportive of that.
> >
> >The issue is hereby re-opened.
> >http://www.w3.org/2001/sw/DataAccess/issues#DESCRIBE
> >
> >I look forward to proposals to close/postpone it. I gather Andy is
> >available to discuss them only thru this Friday.
> >
> >There's still a reasonably good chance that we can get
> >a SPARQL CR out next week, if we come to consensus on
> >something quickly.
> >
> 
> I'm happy with postponing it, either moving it to an appendix or a 
> separate, linked document.
> 
> Either way, it could also link to a more discursive WG Note that surveys 
> the current options.  (There's material out there already so it might be 
> sufficient to just link to it after reviewing it - and then there's service 
> description or different service endpoints.)
> 
> I would like to see the keyword reserved and the current syntax maintained 
> as it is already out there.  I'm neutral as to adding the extra argument as 
> an optional part or the syntax.  ARQ would support it if it were added.
> 
> [[Graham's proposal only captures the "set of known algorithms" usage, not 
> the case where what is returned is influenced by the data itself (e.g. find 
> a resource of type book, so return the ISBN number).  But my feeling is 
> that the "set of known algorithms" is a common usage.]]

To mind mind, data-influenced results is the real value of DESCRIBE.
User-specified results are just exotic templates. The world gets great
use (conventional web browser use) out of letting the content provider
decide what data to return and DESCRIBE gives exactly that access to the
nodes found in a query. Whatever other cool mechanisms we specicy, this
use case is best met by something that is willfully under-specified.

I think leaving DESCRIBE in but sticking <non-normative></non-normative>
around it allows the world to go ahead and play with what might be a
very important development in letting the machine click for you.
-- 
-eric

office: +81.466.49.1170 W3C, Keio Research Institute at SFC,
                        Shonan Fujisawa Campus, Keio University,
                        5322 Endo, Fujisawa, Kanagawa 252-8520
                        JAPAN
        +1.617.258.5741 NE43-344, MIT, Cambridge, MA 02144 USA
cell:   +81.90.6533.3882

(eric@w3.org)
Feel free to forward this message to any list for any purpose other than
email address distribution.