pnowojski commented on a change in pull request #6787: [FLINK-8577][table] Implement proctime DataStream to Table upsert conversion URL: https://github.com/apache/flink/pull/6787#discussion_r249727908
########## File path: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/rules/logical/CalcUpsertToRetractionTransposeRule.scala ########## @@ -0,0 +1,109 @@ +/* + * 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.table.plan.rules.logical + +import org.apache.calcite.plan.RelOptRule.{none, operand} +import org.apache.calcite.plan.{RelOptRule, RelOptRuleCall} +import org.apache.calcite.rex.{RexCall, RexInputRef} +import org.apache.calcite.sql.SqlKind +import org.apache.flink.table.plan.nodes.logical.{FlinkLogicalCalc, FlinkLogicalUpsertToRetraction} + +import scala.collection.JavaConversions._ + +/** + * Use this rule to transpose Calc through UpsertToRetraction relnode. It is beneficial if we get + * smaller state size in upsertToRetraction. + */ +class CalcUpsertToRetractionTransposeRule extends RelOptRule( + operand(classOf[FlinkLogicalCalc], + operand(classOf[FlinkLogicalUpsertToRetraction], none)), + "CalcUpsertToRetractionTransposeRule") { + + override def matches(call: RelOptRuleCall): Boolean = { + val calc = call.rel(0).asInstanceOf[FlinkLogicalCalc] + val upsertToRetraction = call.rel(1).asInstanceOf[FlinkLogicalUpsertToRetraction] + + // column pruning or push Filter down + calc.getRowType.getFieldCount <= upsertToRetraction.getRowType.getFieldCount && Review comment: However I see one more issue here. Does this condition make sense? Shouldn't this be resolved/decided by cost base optimiser? Why do we need this heuristic? What if calc adds column(s) while also being a very selective filter? IMO this condition should be completely dropped and only conditions that guards the correctness should be tested here. Whether `calc` should or shouldn't be pushed down through `UpsertToRetraction` should be decided by the cost of the plan. Cost of `UpsertToRetraction` should reflect both the number of rows and size of the rows. Is this happening? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
