Monday, October 10, 2005

SPARQLing

toward SPARQL CR, PR, REC, press releases... schedule review "...the valueTesting and sort issues continue to act like difficult-to-drain swamps...The objection from Network Inference saying that SPARQL should have used XQuery remains...I wanted to have a much smaller spec (without SOURCE/GRAPH etc.) and finish earlier... since we didn't meet the April CR milestone, I'm pretty much convinced that what I want and what this WG wants are pretty different..."

RDF DAWG Issues List: sort, "...re-opened 2005-08-03 following comment ORDER with IRIs." (also, extending < operator and lexical vs value space), valueTesting, disjunction, countAggregate , "these are complicated in RDF due to open world notions of equality and inequality.", and bnodeRef, "Users seem to find it useful to refer to bnodes given by the server, though the scope of a bnode is usually one lexical graph."

A look at implementing SPARQL using Lisp: "...parts of the grammar weren't designed to be easily parsed; it has a number of features that are clearly user-focused, such as optional dots. (If it were designed to be easily parsed, it would be s-expressions from the start...)...I have seen objections made to UNION, on the grounds that implementation is difficult. I'd like to nip that one in the bud; I wrote twinql in a few weeks, not full-time, inventing algorithms as I went, and UNION wasn't difficult." Alternatively, I guess it could've been XML.

A response to "evaluating SPARQL w.r.t an RDF query language survey" which lists how you can (or cannot) perform queries that: "Return the labels of all topics that are not titles of publications", "Count the number of authors of a publication" and "Return all publications where the page number is the integer value 8."

No comments: