ulysses-you commented on code in PR #57181: URL: https://github.com/apache/spark/pull/57181#discussion_r3576928280
########## sql/core/src/main/scala/org/apache/spark/sql/execution/adaptive/ConvertSortMergeJoinToShuffledHashJoin.scala: ########## @@ -0,0 +1,229 @@ +/* + * 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.spark.sql.execution.adaptive + +import scala.annotation.tailrec + +import org.apache.spark.sql.catalyst.expressions.{Alias, Attribute, CaseWhen, Cast, Coalesce, Expression, If, Literal, Lower, String2TrimExpression, Substring, UnsafeRow, Upper} +import org.apache.spark.sql.catalyst.optimizer.{BuildLeft, BuildRight, BuildSide, JoinSelectionHelper} +import org.apache.spark.sql.catalyst.plans.LeftExistence +import org.apache.spark.sql.catalyst.plans.logical.Join +import org.apache.spark.sql.catalyst.plans.logical.statsEstimation.EstimationUtils +import org.apache.spark.sql.catalyst.rules.Rule +import org.apache.spark.sql.execution.{CollectMetricsExec, FilterExec, ProjectExec, SortExec, SparkPlan} +import org.apache.spark.sql.execution.aggregate.BaseAggregateExec +import org.apache.spark.sql.execution.exchange.{ENSURE_REQUIREMENTS, EnsureRequirements} +import org.apache.spark.sql.execution.joins.{BaseJoinExec, ShuffledHashJoinExec, SortMergeJoinExec} +import org.apache.spark.sql.execution.window.{WindowExecBase, WindowGroupLimitExec} + +/** + * Converts a [[SortMergeJoinExec]] into a [[ShuffledHashJoinExec]] during adaptive execution when + * a build side's materialized shuffle statistics show it is small enough for a local hash map. + * Unlike [[DynamicJoinSelection]], this runs on the physical plan, so it can reach the input Review Comment: Consolidated onto the physical rule, as suggested. `DynamicJoinSelection` is renamed to `DemoteBroadcastHashJoin` and now only emits `NO_BROADCAST_HASH` - its `PREFER_SHUFFLE_HASH`/`SHUFFLE_HASH` selection is gone, and the `PREFER_SHUFFLE_HASH` hint object plus its helpers/consumers are removed entirely, so there is no transitional second path to keep in sync. The physical rule now owns the SHJ-over-SMJ selection with a two-mode `findShuffleStage`: the direct-input-shuffle case is always on (gated only by the master switch `spark.sql.adaptive.convertSortMergeJoinToShuffledHashJoin.enabled`, default `true`), and the look-through-operators capability sits behind `spark.sql.adaptive.convertSortMergeJoinToShuffledHashJoin.lookThroughOperators.enabled` (default `false`). The always-on half reproduces the old default behavior (inert unless `maxShuffledHashJoinLocalMapThreshold` is raised), and only the new look-through capability is opt-in. -- 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]
