Hallo I can donate our facette module to the lucene project.
The implementation relies on field cache only, no index scheme, no cached filters etc. It is small (about 600 lines of code in 10 classes). I didn't measure performance, but it handles 1Mio documents (30GB) without problems. I suppose it might fit the requirements described in LUCENE-3079. The module supports - single valued facets - multi valued facets - facet filters - evaluation of facet values that would dismiss due to other facet filters. Let me explain the last point: For the user a facet query (color==green) AND (shape==circle OR shape==square) may look like Facet color [ ] (3) red [x] (5) green [ ] (7) blue Facet shape [x] (9) circle [ ] (4) line [x] (2) square The red/blue/line facet values will display even though the corresponding documents are not in the result set. Also there is support for filtered facet values with zero results, so users understand why they do not get results. So how to start? Preparing a patch against trunk (currently it is 3.1)? Stefan Trcek --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
