On Thu, Oct 15, 2009 at 11:53 AM, Yonik Seeley <yo...@lucidimagination.com> wrote: > And it seems like a PQ per segment simply delays many of the slow > lookups until the end where the PQs must be merged.
Actually, I'm wrong about that part - one can simply merge on values... there will be lots of string comparisons (and a new merge queue) but no binary searches. Of course a different part of the proposal is problematic: "This way, we avoid the lookup from value->ord, and can keep the simpler API: ScoreDocComparator, which is much easier to extend for custom sorting." We really don't want two different sort API's right? So could you still make this faster using the new comparator APIs? -Yonik http://www.lucidimagination.com --------------------------------------------------------------------- To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org For additional commands, e-mail: java-dev-h...@lucene.apache.org