On 19/07/16 14:44, Paul Hermans wrote:
Side question: Are the SPARQL 1.0 and 1.1 tests suites regularly tested on?
ARQ is SPARQL 1.1 - there
The query in question is:
SELECT * { :x ?p 456. . }
"456." changed from being the decimal number 456. to being integer 456
and a DOT so that makes two DOTs. Illegal.
This was to align with XSD forms (the canonical form of xsd:decimal
changed) and agrees with RDF 1.1 Turtle.
"--syntax=sparql10" will trigger the strict SPARQL 1.0 parser. You
still get SPARQL 1.1 execution semantics.
Tests are maintained by a community group:
https://github.com/w3c/rdf-tests.
Every build runs the tests, or the equivalent of. The tests in ARQ were
donated to W3C as part of the SPARQL 1.1 tests.
Or you can run them locally using the command line "qtest"
See http://sparqlscore.com
Be careful - it isn't accurate and "The soon-to-be reference in
triplestore benckmarking" is a claim, not a fact.
Despite statements to the contrary, ARQ does pass the whole test suite.
But they run the test via Fuseki in some unclear setup.
The GeoSPARQL tests aren't W3C.
The Grid Observatory 3.0 tests are who knows what.
Andy
