The following query:
Gives 6 for ?openTriplets when run at sparql.org on the books database. This query:
Take the query and remove the second triple pattern and the filter and ?openTriplets will have a non-zero answer.
Is the second query exposing a bug in Fuseki/ARQ? Or is there something I don't understand?
Yes this does appear to be a bug in ARQ, a quick bit of digging with the query validator to inspect the different algebras shows that the second query causes ARQ to rename the variables differently (which it does to get scope correct when subqueries are involved).
I filed this as JENA-389, I suspect Andy shall have to take a look at this to figure out why this goes wrong because this is fairly deep in the ARQ engine.
answered 01 Feb '13, 05:01
Rob Vesse ♦