6198 2008-10-31 07:56:13 +0000 [FT] TestSuite - Weight 0 2008-11-07 15:00:05 +0000 1 1 1 Unclassified XPath / XQuery / XSLT Full Text 1.0 Candidate Recommendation All All CLOSED FIXED http://basex.org/ P2 normal --- 1 christian.gruen jim.melton public-qt-comments oldest_to_newest 22308 0 christian.gruen 2008-10-31 07:56:13 +0000 Hi, the last post for today.. The query "FTSelection-Weight-q1b.xq" sets the weight of all results to 0.0. This is why I would expect 0 hits here. Instead, the current test suite file includes results. If results are correct for this query, an implementation would have to make an explicit distinction between false hits and hits with "0.0" as scoring value - this is probably not what implementors and users would expect. Last but not least, if 0.0 scores are valid results, this should at least be implementation dependent. Thanks, Christian, BaseX Team http://www.basex.org 22382 1 jim.melton 2008-11-06 23:29:47 +0000 Christian, your report has merit, but recall that the meaning of weight values is implementation-defined (or is that -dependent?). There is nothing in the spec that says that a weight of 0.0 means "ignore this part of the query", so the assumption that the query should return nothing is mistaken. We all believe, of course, that a query term with a weight of 0.0 should be considered "less relevant" or "less important" than a query term with a weight of 999.999, but that's not guaranteed by the spec. Nonetheless, it's probably appropriate to give two alternate possible results -- one as it's already specified, but the other as "empty result". It is perfectly reasonable, even though not required, that an implementation would choose to interpret 0.0 to mean "completely irrelevant". Thanks for the report. I'll go ahead and add that second alternative result, so I'm marking this bug as FIXED. If you agree with this resolution, please mark the bug CLOSED. 22386 2 christian.gruen 2008-11-07 15:00:05 +0000 Dear Jim, thank you for the detailed comments. Yes, an alternative result surely is the best solution here to consider different implementations. I've closed this bug. Christian, BaseX Team http://www.basex.org