Stefan Manegold wrote:
Log Message: after running the testset of INEX efficiency track, i found and fixed several small bugs that caused queries to fail.
Would it be an option to consider adding these tests to the standard testset to be run (and checked) automatically every night?
Stefan
well, the test set is indeed more representative for pf/tijah than our own test cases. however, it is usually run on a large test corpus (XML version of Wikipedia 2006, ~5GB). in order to check that the queries don't result in error messages it is probably possible to just index a tiny fraction of this corpus, however most queries will result in empty result sets then. furthermore, it would take some manual work to select all "structural different" queries out of the 568 queries. many of them would in fact test the same functionality. so, it is possible, but with some work to do. -henning