I can confirm that, just had a chat at RDF4JUsers https://groups.google.com/forum/#%21topic/rdf4j-users/p-uqdjgGMHs where I was told that:

"In terms of getting the query to parse, why not just reorder the BIND() to the front of the group? This is what blazegraph is doing internally. "

So you guessed right (and we are going to treat those queries the same way).

Thanks and greetings,

Adrian

Hi!

On 1/14/17 9:23 AM, Adrian Bielefeldt wrote:

...

I was wondering if anybody could help me understand why (and how) this query:

SELECT DISTINCT ?item WHERE { ?tree0 wdt:P31 ?item . BIND (wd:Q146 AS ?tree0) }

is being accepted by query.wikidata.org.

Probably Blazegraph engine considers it analogous with the query having the bind in front, or maybe optimization reorders it.

...

At the moment we are parsing them using OpenRDF, which fails (as far

as I can see understandably) at parsing this query.

Well, the standard does say you can not use a variable in BIND that was previously used, and I guess OpenRDF parser chooses to be more strict than Blazegraph.