sorry for the second email in a row...
I've hit another change in the behavior of the VALUES clause. In this case, there is a change in how VALUES interacts with FILTER inside an OPTIONAL group.
Please consider the following query:
This query returns a single solution with
datasetBound set to true.
I'm pretty sure that in previous versions the results would have been false.
Actually, I was able to obtain this result by wrapping the VALUES section within a pair of braces
Looking at specs (https://www.w3.org/TR/sparql11-query/#sparqlAlgebraFinalValues
), I'm unsure which behavior is right. My idea in favor of the new behavior is that the VALUES section is "evaluated" after the FILTER, thus joining an empty binding set (because FILTER is unsatisfied) with the contents of VALUES.