zacharymorn commented on code in PR #1018:
URL: https://github.com/apache/lucene/pull/1018#discussion_r922765220


##########
lucene/core/src/java/org/apache/lucene/search/BooleanWeight.java:
##########
@@ -191,6 +191,69 @@ public long cost() {
   // or null if it is not applicable
   // pkg-private for forcing use of BooleanScorer in tests
   BulkScorer optionalBulkScorer(LeafReaderContext context) throws IOException {
+    if (scoreMode == ScoreMode.TOP_SCORES) {
+      if (query.getMinimumNumberShouldMatch() > 1 || weightedClauses.size() > 
2) {
+        return null;
+      }
+
+      List<ScorerSupplier> optional = new ArrayList<>();
+      for (WeightedBooleanClause wc : weightedClauses) {
+        Weight w = wc.weight;
+        BooleanClause c = wc.clause;
+        if (c.getOccur() != Occur.SHOULD) {
+          continue;
+        }
+        ScorerSupplier scorer = w.scorerSupplier(context);
+        if (scorer != null) {
+          optional.add(scorer);
+        }
+      }
+
+      if (optional.size() <= 1) {
+        return null;
+      }
+
+      List<Scorer> optionalScorers = new ArrayList<>();
+      for (ScorerSupplier ss : optional) {
+        optionalScorers.add(ss.get(Long.MAX_VALUE));
+      }
+
+      return new BulkScorer() {

Review Comment:
   > It shouldn't be slower than the current code in main since main is using 
DefaultBulkScorer, is it?
   
   The baseline of all of the above benchmark results are still using the head 
prior to all BMM changes. Since this approach (anonymous bulk scorer + BMM 
scorer) still has similar performance boost with the previous one (just BMM 
scorer) for top-level disjunctions, but no impact to nested boolean queries, I 
would think so? I'm not sure I'm fully understanding this question though.



-- 
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

Reply via email to