> From: Dmitry Serebrennikov [mailto:[EMAIL PROTECTED]] > > > > It should not be the case. The collect() method should be > > called at most once per document. > > > This is a good news! This would make counting that much more > efficient. > My main concern was in the BooleanScorer, and I just verified > that I was > worried needlessly - it maintains its own hashtable to avoid double > counting. On a related issue, are there any guarantees about > the order > of document numbers in the calls to collect()?
Calls are always in order of increasing document number. Doug