Re: XML serialization of SPARQL

On Wed, Mar 16, 2005 at 04:04:35PM +0000, Seaborne, Andy wrote:
> Kendall Clark wrote:

> SPARQL/QL is defined abstractly (basic pattern, group, union, optional, 
> graph,
> filter)

And except for a few exceptions (concessions, really, to *some* sense
of human readability), that's pretty much the level of my XML
serialization. I'm ignoring all the syntactic sugar, as I don't really
care about roundtripping.

> Could you say what connections and dependencies you are discovering?

Well, for example, WSDL defines several possible ways of passing
messages around (in the Bindings spec), only one of which is
"serialize into a GET" (which is, basically, a REST fetishization of
the URI, treating it *precisely* as a message body -- but that's no
matter :>); and another of which is to POST or otherwise pass an XML
instance that represents the message. So I suspect this XML
serialization of SPARQL queries should "play nicely" with WSDL's sense
of how to pass XML instances around.

So, hmm, I've not really discovered anything earth-shattering yet, I
just kinda expect to find some bits here and there that need to be
synched.

> Just asking but is it proposed that a SPARQL/XML syntax be part of our Last 
> Call
> bundle of docs?  Or is it a WG note?  Something else?

I can't imagine it could be done in time to go to Last Call w/ the
query doc. As I've said, I'm fine with the query doc going to LC
before the protocol doc. I think it would be good if SparqlX could go
to LC at the same time as protocol.

I'd like it to be a normative way of representing a query, which is to
say: no processor would be required to consume it, but it would be
*the* XML serialization of a SPARQL query.

> Should there be an RDF encoding of the query structure? 

Sure, why not? I don't have a problem with that. I'm unlikely to do
the work, but I'd support someone doing it.

> If we want 
> multiple syntaxes, a shared abstract model of a query encoded into various 
> forms would be cool.

Yes, and I think such a thing lives in rq23. Multiple "major syntaxes"
(i.e., human readable, XML, RDF) would serve as a useful kind of
debugging/reality-check, IMO.

Kendall

Received on Wednesday, 16 March 2005 16:20:15 UTC