Re: Do we have consensus that we don't need more R2RML syntaxes?

On 21 Jun 2011, at 19:42, David McNeil wrote:
>> Well, you need to define conformance somehow. If my R2RML editor can only produce, say, RDFa, is it conforming to R2RML? If your R2RML processor only understands, say, RDF/XML, is it conforming to R2RML? And do they interoperate?
>> 
>> Interoperability requires shared syntax. RDF is not a syntax, Turtle is.
> 
> Right. I would say that R2RML is represented as RDF and point to the well known serialization formats as alternate implementations. We could point out that an R2RML implementation does not have to support all of them.

If an implementation that only supports Turtle and an implementation that only supports RDF/XML can both claim to be conforming, then we have failed.

> We could also point out that R2RML does not address conversion between formats which strikes me as a solved problem.

It will be a solved problem when the average user doesn't need to download and install anything to convert between RDF formats. Until then, I'd rather deal with the issue in the spec.

> In terms of interoperability I think the use of vendor specific SQL in a mapping is much more significant than the RDF serialization format used.

That is true. So why are you so concerned about the RDF serialization format that's being used?

You have not actually presented any argument why you are opposed to requiring Turtle support in conforming implementations.

Richard

Received on Tuesday, 21 June 2011 20:21:22 UTC