>> Grouping etc could work solely on a docvalues api and maybe they have >> a special ctor to take a SortedBytes impl (and fieldcache implements >> the docvalues api or some other backwards mechanism). >> This would remove the redundant implementations (both fieldcache and >> DV implementations) in stuff like grouping, join, sort, etc +1 That is great, b/c there already are docvalues based grouping implementations.
We don't need to have a special SortedBytes ctor, right? The grouping ctors do take a DocValues.Type (joining, faceting etc can do the same). In the setNextReader method IR.docValues(groupField) is invoked. I'll also close LUCENE-3360 since it only couples the FC with IR. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org