COUNT(DISTINCT *) can generate incorrect results.

Using COUNT(DISTINCT *) appears to generate the wrong result, in that it appears to include system generated bindings as part of the solution when DISTINCTing them. (I think similar to the issue in JENA-211).

For example, I'd expect these 2 queries to return the same results, as ?lbl should be the only variable considered part of the solutions.

SELECT (COUNT(DISTINCT *) as ?c)
WHERE { <child3> <p>+/<label> ?lbl }

SELECT (COUNT(DISTINCT ?lbl) as ?c)
WHERE  { <child3> <p>+/<label> ?lbl }

However these return different results. A full simple repro case is available at https://github.com/superfell/jena-count-distinct-star