Re: SPARQL Protocol question - details of 2.1.2 query In Message

Hi Ray,

Again, apologies for the delay getting back to you. We've updated the 
copy of the XML schema at the namespace URI to be the latest version. 
Please let me know if there are any problems.

Lee

Hookway, Ray wrote:
> I found what I  believe to be a problem in the SPARQL protocol
> specification. The basic problem is a disparity between
> http://www.w3.org/2005/09/sparql-protocol-types
> <http://www.w3.org/2005/09/sparql-protocol-types>  which is imported by
> http://www.w3.org/TR/rdf-sparql-protocol/sparql-protocol-query.wsdl
> <http://www.w3.org/TR/rdf-sparql-protocol/sparql-protocol-query.wsdl>
> and http://www.w3.org/TR/rdf-sparql-protocol/sparql-protocol-types.xsd
> <http://www.w3.org/TR/rdf-sparql-protocol/sparql-protocol-types.xsd> .
> The later two are both referenced in the spec and
> http://www.w3.org/TR/rdf-sparql-protocol/sparql-protocol-types.xsd
> <http://www.w3.org/TR/rdf-sparql-protocol/sparql-protocol-types.xsd>  is
> quoted in the spec. The version of sparql-protocol-types which is
> referenced by the wsdl document describes the type of a query-request as
> <xs:all>. The other version describes the type of a query-request as an
> <xs:sequence>. The former allows query, default-graph-uris and
> named-graph-uris in any order. However, it contains what looks like an
> error in declaring a maxOccurs of "unbounded" for default-graphs-uri and
> named-graph-uri. MaxOccurs can only be 0 or 1 for elements of <xs:all>.
> The intent seems to be to allow multiple default-graph-uris and
> named-graph-uris, but this isn't allowed by XMLSchema. It looks to me
> like you're caught in a bind here. <xs:sequence> requires a fixed order,
> but allows unbounded occurrences. <xs:all> allows arbitrary order, but
> requires at most one occurrence. (All this has to do with the XML
> description of the message. I'm not sure how this gets mapped by the
> HTML binding - i.e., the binding may allow a rearrangement of the
> elements although I doubt it.)
>  
> Please let me know if this is not clear.
>  
> Thanks.
>  
> -Ray Hookway
> Hewlett-Packard
>

Received on Friday, 28 September 2007 13:48:08 UTC