somu-imply commented on code in PR #14408: URL: https://github.com/apache/druid/pull/14408#discussion_r1305704388
########## processing/src/main/java/org/apache/druid/query/aggregation/first/NumericFirstVectorAggregator.java: ########## @@ -0,0 +1,174 @@ +/* + * 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.druid.query.aggregation.first; + +import org.apache.druid.common.config.NullHandling; +import org.apache.druid.query.aggregation.VectorAggregator; +import org.apache.druid.segment.vector.VectorValueSelector; + +import javax.annotation.Nullable; +import java.nio.ByteBuffer; + +/** + * Class for vectorized version of first/earliest aggregator over numeric types + */ +public abstract class NumericFirstVectorAggregator implements VectorAggregator +{ + static final int NULL_OFFSET = Long.BYTES; + static final int VALUE_OFFSET = NULL_OFFSET + Byte.BYTES; + final VectorValueSelector valueSelector; + private final boolean useDefault = NullHandling.replaceWithDefault(); + private final VectorValueSelector timeSelector; + private long firstTime; + + public NumericFirstVectorAggregator(VectorValueSelector timeSelector, VectorValueSelector valueSelector) + { + this.timeSelector = timeSelector; + this.valueSelector = valueSelector; + firstTime = Long.MAX_VALUE; + } + + @Override + public void init(ByteBuffer buf, int position) + { + buf.putLong(position, Long.MAX_VALUE); + buf.put(position + NULL_OFFSET, useDefault ? NullHandling.IS_NOT_NULL_BYTE : NullHandling.IS_NULL_BYTE); + initValue(buf, position + VALUE_OFFSET); + } + + @Override + public void aggregate(ByteBuffer buf, int position, int startRow, int endRow) + { + final long[] timeVector = timeSelector.getLongVector(); + final boolean[] nullTimeVector = timeSelector.getNullVector(); + final boolean[] nullValueVector = valueSelector.getNullVector(); + firstTime = buf.getLong(position); + + // the time vector is already sorted + // if earliest is on the default time dimension + // but if earliest uses earliest_by it might use a secondary timestamp + // which is not sorted. For correctness, we need to go over all elements. + // A possible optimization here is to have 2 paths one for earliest where + // we can take advantage of the sorted nature of time + // and the earliest_by where we have to go over all elements. + int index; + + for (int i = startRow; i < endRow; i++) { + index = i; + if (nullTimeVector != null && nullTimeVector[index]) { + continue; Review Comment: So currently in case of non-null timestamps, we just take the first value. We iterate over the values and only update the first time if the current time is less than the earliest time. The issue arises when all timestamps are null. Considering the earliest where the time selector is __time, the chances of this happening are more when used on a secondary timestamp through earliest_by. In such a case, should we even return any results ? The docs point that ``` If expr comes from a relation with a timestamp column (like __time in a Druid datasource), the "earliest" is taken from the row with the overall earliest non-null value of the timestamp column. ``` -- 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]
