[
https://issues.apache.org/jira/browse/FLINK-4937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15682381#comment-15682381
]
ASF GitHub Bot commented on FLINK-4937:
---------------------------------------
Github user sunjincheng121 commented on a diff in the pull request:
https://github.com/apache/flink/pull/2792#discussion_r88824600
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/api/table/runtime/aggregate/IncrementalAggregateReduceFunction.scala
---
@@ -0,0 +1,69 @@
+/*
+ * 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.flink.api.table.runtime.aggregate
+
+import org.apache.flink.api.common.functions.ReduceFunction
+import org.apache.flink.api.table.Row
+import org.apache.flink.util.Preconditions
+
+/**
+ * For Incremental intermediate aggregate Rows, merge every row into
aggregate buffer.
+ *
+ * @param aggregates The aggregate functions.
+ * @param groupKeysMapping The index mapping of group keys between
intermediate aggregate Row
+ * and output Row.
+ */
+class IncrementalAggregateReduceFunction(
+ private val aggregates: Array[Aggregate[_]],
+ private val groupKeysMapping: Array[(Int, Int)],
+ private val intermediateRowArity: Int)extends ReduceFunction[Row] {
+
+ Preconditions.checkNotNull(aggregates)
+ Preconditions.checkNotNull(groupKeysMapping)
+ @transient var accumulatorRow:Row = _
+
+ /**
+ * For Incremental intermediate aggregate Rows, merge value1 and value2
+ * into aggregate buffer, return aggregate buffer.
+ *
+ * @param value1 The first value to combined.
+ * @param value2 The second value to combined.
+ * @return The combined value of both input values.
+ *
+ */
+ override def reduce(value1: Row, value2: Row): Row = {
+
+ if(null == accumulatorRow){
+ accumulatorRow = new Row(intermediateRowArity)
+ }
+
+ // Initiate intermediate aggregate value.
+ aggregates.foreach(_.initiate(accumulatorRow))
--- End diff --
Because of last week's vacation, I'm sorry to reply back later. Thanks to
@fhueske and @wuchong 's discussion. At the beginning I did try to return
value1 or value2 , but it does not work, @fhueske is right! So I added the
accumulatorRow to compute. At the beginning I created a new accumulatorRow
object every time in the reduce method, and later the accumulatorRow was
changed to a class-shared variable for reduced create object, which resulted in
this error. and so before optimizing ReducingState maybe we need to create
accumulatorRow each time in reduce method And after optimizing ReducingState,
we can remove accumulatorRow. is this right?
> Add incremental group window aggregation for streaming Table API
> ----------------------------------------------------------------
>
> Key: FLINK-4937
> URL: https://issues.apache.org/jira/browse/FLINK-4937
> Project: Flink
> Issue Type: Sub-task
> Components: Table API & SQL
> Affects Versions: 1.2.0
> Reporter: Fabian Hueske
> Assignee: sunjincheng
>
> Group-window aggregates for streaming tables are currently not done in an
> incremental fashion. This means that the window collects all records and
> performs the aggregation when the window is closed instead of eagerly
> updating a partial aggregate for every added record. Since records are
> buffered, non-incremental aggregation requires more storage space than
> incremental aggregation.
> The DataStream API which is used under the hood of the streaming Table API
> features [incremental
> aggregation|https://ci.apache.org/projects/flink/flink-docs-release-1.2/dev/windows.html#windowfunction-with-incremental-aggregation]
> using a {{ReduceFunction}}.
> We should add support for incremental aggregation in group-windows.
> This is a follow-up task of FLINK-4691.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
