jpountz commented on PR #13773: URL: https://github.com/apache/lucene/pull/13773#issuecomment-2346309641
TopScoreDocCollectorManager is supposed to return the top-N results that have the best score, tie-broken by doc ID. So (docID=0, score=1) compares better than (docID=155, score=1) and should appear first in the resulting top docs. > I think that we are still early terminating though or you mean that we shouldn't have as there were docs with same score but lower doc ids? Right. Looking at the code, it looks like the problem is here: https://github.com/apache/lucene/blob/ff8b81afc59dc09ea4dfdd02a58398f08fcad8a8/lucene/core/src/java/org/apache/lucene/search/TopScoreDocCollector.java#L260. If a segment has two partitions, then its leaf collectors will collect hits with the same "docBase", so the tie-breaking logic will not know that hits from the first partition should compare better than hits from the second partition. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
