> From: Dave <[email protected]> > To: oslc-core <[email protected]> > Date: 05/11/2010 10:18 AM > Subject: [oslc-core] Should RDF/XML be MUST? > Sent by: [email protected] > > Sorry to raise this old issue again, but I've been getting some new > feedback that the Core spec should not be so prescriptive (or is it > proscriptive) about RDF/XML representation. I captured this feedback > in a new issue on the issues page: > > OPEN Consensus among RDF experts seems to be that RDF/XML is not the > best representation for RDF, so why do we mandate it as a MUST in the > Core spec. In reality, most OSLC workgroups will probably make RDF/XML > a MUST, but perhaps we should leave that up to them. Here are two > alternatives: (DaveJohnson, 05/11/2010) > * Option #1 - say this: OSLC services SHOULD provide RDF/XML > representations for all resources and MAY provide Turtle, JSON or Atom > representations. > * Option #2 - say this: OSLC services SHOULD provide an RDF > serialization, either RDF/XML or Turtle, and MAY provide JSON or Atom > representations. > * *Response* pending... (DaveJohnson 05/11/2010) > > As always, feedback, comments, etc. are most welcome. >
I think the core requirement to capture is: if a domain spec defines a resource, it should use RDF to define that resource and its properties. Since there exists already standardized ways to serialize this resource to XML, then we should leverage that standards (RDF/XML). There could be cases where XML-only format may be valid, where full valid RDF/XML may not be warranted. How do you capture this in the spec? Perhaps: Option #3 - say this: OSLC services MUST provide a RDF/XML valid XML representation for OSLC defined resources and MAY provide Turtle, JSON or Atom representations. This leaves it open for "non-OSLC defined resources". Then domain specs can mandate it further if needed as well. Thanks, Steve Speicher | IBM Rational Software | (919) 254-0645
