Re: SPARQL Protocol possible schema tweaks (sparql-protocol-types.xsd)

On Mon, Jun 06, 2005 at 09:29:59AM -0400, Lee Feigenbaum wrote:
> 
> Hi everyone,
> 
> Looking over the XML schema referenced in the May 27 working 
> draft of the protocol specification, I had a couple  of (very) small 
> comments:

Hi Lee, thanks for the comments. Responses interspersed:

> 1) The text of the specification indicates that an RDF Dataset 
> must contain one default graph, whereas the rdf-dataset element 
> (and accompanying inline documentation) makes the default
> graph optional. I *think* the latter is intended to be the case, in
> which case the text should be updated to reflect this.

Ah, yes, good catch. These cardinality constraints are still up in the air,
but this is a straight bug.

> 1b) In answer to the parenthetical question "can there be zero
> datasets?", I'd guess that at this point in time the answer to
> this question is "yes", given the decision to allow graphs to
> be specified either as datasets in the protocol or via FROM 
> and FROM NAMED in the query itself?

Yes, indeed.

> 2) Unless I'm mistaken, I believe that the minOccurs="0" on the
> vbr:sparql and rdf:RDF elements within the choice group of the
> query-result element mean that it is valid for a query response to have 
> an empty query-result element. Is this purposeful? It doesn't seem to
> fit with the text of the specification ("either one or the other of two 
> further elements"), and I'm not sure what it would mean semantically,
> either.

Hmm, this is also just a straight bug.

Thanks a lot Lee.

Kendall Clark

Received on Monday, 6 June 2005 14:07:54 UTC