flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-9713) Support versioned joins in planning phase
Date Thu, 09 Aug 2018 10:14:01 GMT

    [ https://issues.apache.org/jira/browse/FLINK-9713?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16574654#comment-16574654
] 

ASF GitHub Bot commented on FLINK-9713:
---------------------------------------

twalthr commented on a change in pull request #6299: [FLINK-9713][table][sql] Support processing
time versioned joins
URL: https://github.com/apache/flink/pull/6299#discussion_r208876405
 
 

 ##########
 File path: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/rules/logical/LogicalCorrelateToVersionedJoinRule.scala
 ##########
 @@ -0,0 +1,243 @@
+/*
+ * 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 java.util
+import java.util.Collections
+
+import org.apache.calcite.plan.RelOptRule.{any, none, operand, some}
+import org.apache.calcite.plan.{RelOptRule, RelOptRuleCall}
+import org.apache.calcite.rel.RelNode
+import org.apache.calcite.rel.core.TableFunctionScan
+import org.apache.calcite.rel.logical.LogicalCorrelate
+import org.apache.calcite.rex._
+import org.apache.calcite.sql.fun.SqlCastFunction
+import org.apache.flink.table.api.{Table, ValidationException}
+import org.apache.flink.table.functions.TableVersionFunction
+import org.apache.flink.table.functions.sql.ProctimeSqlFunction
+import org.apache.flink.table.functions.utils.TableSqlFunction
+import org.apache.flink.table.plan.logical.rel.LogicalVersionedJoin
+import org.apache.flink.table.plan.schema.TimeIndicatorRelDataType
+import org.apache.flink.table.plan.util.RexDefaultVisitor
+import org.apache.flink.table.typeutils.TypeStringUtils
+import org.apache.flink.util.Preconditions.checkState
+
+class LogicalCorrelateToVersionedJoinRule
+  extends RelOptRule(
+    operand(classOf[LogicalCorrelate],
+      some(
+        operand(classOf[RelNode], any()),
+        operand(classOf[TableFunctionScan], none()))),
+    "LogicalCorrelateToVersionedJoinRule") {
+
+  override def onMatch(call: RelOptRuleCall): Unit = {
+    val logicalCorrelate: LogicalCorrelate = call.rel(0)
+    val leftNode: RelNode = call.rel(1)
+    val rightTableFunctionScan: TableFunctionScan = call.rel(2)
+
+    val cluster = logicalCorrelate.getCluster
+
+    new GetTableVersionFunctionCall(cluster.getRexBuilder, leftNode)
+      .visit(rightTableFunctionScan.getCall) match {
+      case Some(rightVersionedTableCall) =>
+        val underlyingTableHistory: Table = rightVersionedTableCall.tableVersionFunction.table
+        val relBuilder = this.relBuilderFactory.create(
+          cluster,
+          underlyingTableHistory.relBuilder.getRelOptSchema)
+        val rexBuilder = cluster.getRexBuilder
+
+        val rightNode: RelNode = underlyingTableHistory.logicalPlan.toRelNode(relBuilder)
 
 Review comment:
   Creating a logical plan and inserting it during a phase of the optimization is dangerous.
You would skip steps that need to be applied before. In our case these steps are shown in
`org.apache.flink.table.api.StreamTableEnvironment#optimize`: 
   - convert sub-queries before query decorrelation
   - convert table references
   - decorrelate
   - convert time indicators
   
   It might be better to execute this rule in the "convert table references" step.
   
   Btw in this stage time indicators have not been materialized. So you can also remove the
`CAST` special case. Otherwise I would also recommend changes in `org.apache.flink.table.calcite.RexTimeIndicatorMaterializer#visitCall`.

----------------------------------------------------------------
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:
users@infra.apache.org


> Support versioned joins in planning phase
> -----------------------------------------
>
>                 Key: FLINK-9713
>                 URL: https://issues.apache.org/jira/browse/FLINK-9713
>             Project: Flink
>          Issue Type: Sub-task
>          Components: Table API &amp; SQL
>    Affects Versions: 1.5.0
>            Reporter: Piotr Nowojski
>            Assignee: Piotr Nowojski
>            Priority: Major
>              Labels: pull-request-available
>
> Queries like:
> {code:java}
> SELECT 
>   o.amount * r.rate 
> FROM 
>   Orders AS o, 
>   LATERAL TABLE (Rates(o.rowtime)) AS r 
> WHERE o.currency = r.currency{code}
> should evaluate to valid plan with versioned joins plan node.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message