maedhroz commented on code in PR #2409: URL: https://github.com/apache/cassandra/pull/2409#discussion_r1230131645
########## src/java/org/apache/cassandra/index/sai/disk/v1/bbtree/IntersectingPointValues.java: ########## @@ -0,0 +1,118 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.cassandra.index.sai.disk.v1.bbtree; + +import java.io.IOException; + +import org.apache.lucene.codecs.MutablePointValues; +import org.apache.lucene.index.PointValues; +import org.apache.lucene.util.BytesRef; + +public abstract class IntersectingPointValues extends MutablePointValues +{ + private static final byte[] EMPTY = new byte[0]; + + public boolean needsSorting() + { + return size() > 1; + } + + public abstract void intersect(IntersectVisitor visitor) throws IOException; + + @Override + public int getDocCount() + { + throw new UnsupportedOperationException(); + } + + @Override + public long size() + { + return 1; + } + + @Override + public void getValue(int i, BytesRef packedValue) + { + // no-op + } + + @Override + public byte getByteAt(int i, int k) + { + return 0; Review Comment: > One option here would be to change how we build the in-memory numerics such that they are naturally sorted. I am thinking of a trie here. I like this, and I like it a lot. I'm +100 on just replacing `BlockBalancedTreeRamBuffer` entirely w/ an in-memory trie and blowing away all the `PointValues` garbage. Then we end up having something more like `ImmutableIntersectingPointValues` for all segment building cases? What do we lose if we toss out `BlockBalancedTreeRamBuffer`? `PackedLongValues` packing row IDs more efficiently than the trie stores them as payloads? If losing that and gaining value prefix compression was about break-even in terms of heap used, being able to avoid sorting would still be a win... -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
