I've recently discovered a subtle bug in my SPARQL optimisation which means queries of the following form may fail unexpectedly or not return correct results:

SELECT * WHERE {?s ?p ?o . FILTER(...)} LIMIT 10

This only applies to queries where there is a LIMIT clause (and no DISTINCT/GROUP BY/HAVING/ORDER BY) so should affect a small number of queries. If running such a query in code you can disable this optimisation by setting Options.AlgebraOptimisation to false. The latest nightly builds found in the bin\ directory in the dotNetRDF SVN Repository have a fix for this.

Currently the fix in SVN is to disable such optimisations but we hope to reinstate this optimisation in the next few days once we work out a way to eliminate the problem completely.

Tags: Bug, Bugs, Fix, Issues, Known Issues, Optimisation, SPARQL