antonha commented on code in PR #13149:
URL: https://github.com/apache/lucene/pull/13149#discussion_r1515631535


##########
lucene/core/src/java/org/apache/lucene/util/bkd/DocIdsWriter.java:
##########
@@ -36,6 +36,66 @@ final class DocIdsWriter {
 
   private final int[] scratch;
 
+  private final ScratchDocIdSetIterator scratchDocIdSetIterator = new 
ScratchDocIdSetIterator();
+
+  /**
+   * DocIdSetIterator to be used to iterate over the scratch buffer. A single 
instance is reused to
+   * avoid re-allocating the object. The reset method should be called before 
each use with the
+   * count.
+   *
+   * <p>The main reason for existing is to be able to call the {@link
+   * IntersectVisitor#visit(DocIdSetIterator)} method rather than the {@link
+   * IntersectVisitor#visit(int)} method. This seems to make a difference in 
performance, probably
+   * due to fewer virtual calls then happening (once per read call rather than 
once per doc).
+   */
+  private class ScratchDocIdSetIterator extends DocIdSetIterator {
+
+    private int index = -1;
+    private int count = -1;
+
+    @Override
+    public int docID() {
+      if (index < 0) {
+        return -1;
+      }
+      if (index >= count) {
+        return NO_MORE_DOCS;
+      }
+      return scratch[index];
+    }
+
+    @Override
+    public int nextDoc() throws IOException {
+      index++;
+      if (index >= count) {
+        return NO_MORE_DOCS;
+      }
+      return scratch[index];
+    }
+
+    @Override
+    public int advance(int target) throws IOException {
+      while (index < count && scratch[index] < target) {

Review Comment:
   Maybe we don't need to - but I don't see why this would be scarier than 
advancing using the `nextDoc` method? But I might be missing something - if 
you'd like that change, I'll make it. It might be that a DISI is not the right 
interface for visiting docids within the BKD tree - but I'm guessing changing 
that would be larger and more controversial. 
   
   For large segments the buffer size would almost always be 512 - but there 
would be a lot of buffers. 



-- 
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