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

On 22 Jun 2011, at 03:28, David McNeil wrote:
> Although I received numerous comments at Semtech regarding the verbosity of the Turtle representation and a desire for a more concise syntax

Turtle can be partially blamed for that, but another part is just R2RML's design. It is a lot more verbose than D2RQ's mapping language, despite both being Turtle.

>> [[
>> Conforming implementations MAY support other RDF serializations besides Turtle.
>> ]]
> Would adding that sentence satisfy the goal of the charter to provide an XML representation?

My honest technical assessment: no, it doesn't.

Processing arbitrary RDF/XML with XML tools doesn't work as expected, in general. To make R2RML processable with XML tools, it would need a restricted RDF/XML serialization that prescribes how to deal with nesting, RDF/XML striping, rdf:Description versus class elements and so on. This is what was done for RSS 1.0, for example. It tends to make neither XML users nor RDF users very happy.

Also, sorry but it has to be said, RDF/XML is a steaming pile of shit. The other syntaxes are superior in just about every way, and I'd rather not do anything to further promote the use of RDF/XML.

Adding more syntaxes has many subtle costs, especially with regards to education and uptake. For example, users new to the technology have to make a syntax choice without being in a position of sufficient knowledge to do so; and when they read tutorials or google for examples then they might find different syntaxes, which might be a rather confusing experience for them. In my eyes, that's a strong and compelling argument against XML.

Personally, I'd really prefer not to go there, and just not deliver on the “SHOULD have XML representation” goal in the charter. Having a single normative syntax is best.

(The only thing that the WG has formally decided so far is that Turtle would be used as the syntax in the first public working draft. The option of doing an XML syntax was still very much on the table at that time. AFAIR, syntax hasn't been discussed at all since that time.)


Received on Wednesday, 22 June 2011 09:33:19 UTC