ISSUE-57: Ok to go to Last Call with this text?

Souri, Ashok,

You have threatened a formal objection over ISSUE-57, regarding the requirement for Turtle support. In the previous call, we discussed adding a note to the spec that highlights this as an unresolved question, so that we can go to LC even though there is no consensus.

I've added a big “issue box” on ISSUE-57 to the R2RML spec:

Its full text is quoted below. I have tried to fairly present both viewpoints, but no idea if I succeeded.

Is this acceptable for LC? If not, can you please suggest alternative wording?

I hope that we can remove this LC blocker by accepting the following in the next call:

PROPOSAL: Lack of consensus on the syntax question is noted in the issue box in Section 4.2, which presents both options. A decision on the question will be postponed until after Last Call. This closes ISSUE-57.


ISSUE-57: Should R2RML require a specific syntax?

The working group has not reached consensus on this question.

The view expressed in this Working Draft is that an R2RML mapping document must be a Turtle document, and R2RML processors must support Turtle to be able to read such documents (and may support additional other formats). The advantage of this approach is that it promotes interoperability between different producers and consumers of R2RML files by requiring all to support at least one shared syntax. Without such a shared syntax, an R2RML file created in one tool may be rejected by another tool because both assume different RDF syntaxes. R2RML examples found in educational material may not work in actual implementations due to different syntaxes. This is seen as an impediment to the uptake of R2RML.

A part of the working group holds the conflicting view that it is sufficient to define R2RML in terms of the abstract mapping graph. The specification should not talk about syntax at all, nor impose conformance criteria on syntax support. The definition of “R2RML mapping document” is unnecessary. The advantage of this approach is that R2RML would not be unnecessarily tied to a specific syntax, and it would add flexibility as different syntax flavours of R2RML could be easily defined. It is in the spirit of RDF as an abstract format. Users may have to convert between different RDF syntaxes in order to use R2RML files, but such conversion is not difficult and therefore not seen as an impediment.

There is consensus that Turtle should be used for the examples in this document, as well as for the test cases.

The working group seeks comments and opinions on this question and encourages reports to public-rdb2rdf-comments mailing list.

Received on Friday, 26 August 2011 16:15:27 UTC