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-6232) Support proctime inner equi-join between two streams in the SQL API
Date Thu, 13 Jul 2017 02:06:00 GMT

    [ https://issues.apache.org/jira/browse/FLINK-6232?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16085046#comment-16085046
] 

ASF GitHub Bot commented on FLINK-6232:
---------------------------------------

Github user wuchong commented on a diff in the pull request:

    https://github.com/apache/flink/pull/3715#discussion_r127113760
  
    --- Diff: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/rules/datastream/DataStreamWindowJoinRule.scala
---
    @@ -20,28 +20,60 @@ package org.apache.flink.table.plan.rules.datastream
     
     import org.apache.calcite.plan.{RelOptRule, RelOptRuleCall, RelTraitSet}
     import org.apache.calcite.rel.RelNode
    +import org.apache.calcite.rel.`type`.RelDataType
     import org.apache.calcite.rel.convert.ConverterRule
    +import org.apache.calcite.rex.RexNode
    +import org.apache.flink.table.api.{TableConfig, TableException}
     import org.apache.flink.table.plan.nodes.FlinkConventions
    -import org.apache.flink.table.plan.nodes.datastream.DataStreamRowStreamJoin
    +import org.apache.flink.table.plan.nodes.datastream.DataStreamWindowJoin
     import org.apache.flink.table.plan.nodes.logical.FlinkLogicalJoin
     import org.apache.flink.table.plan.schema.RowSchema
    -import org.apache.flink.table.runtime.join.JoinUtil
    +import org.apache.flink.table.runtime.join.WindowJoinUtil
     
    -class DataStreamRowStreamJoinRule
    +class DataStreamWindowJoinRule
       extends ConverterRule(
           classOf[FlinkLogicalJoin],
           FlinkConventions.LOGICAL,
           FlinkConventions.DATASTREAM,
           "DataStreamJoinRule") {
     
    +  /** Time indicator type **/
    +  private var timeType: RelDataType = _
    +
    +  /** left input lower boudary **/
    +  private var leftLowerBoundary: Long = _
    +
    +  /** left input upper boudary **/
    +  private var leftUpperBoundary: Long = _
    +
    +  /** remain join condition exclude equal condition and time condition **/
    +  private var remainCondition: Option[RexNode] = _
    +
       override def matches(call: RelOptRuleCall): Boolean = {
         val join: FlinkLogicalJoin = call.rel(0).asInstanceOf[FlinkLogicalJoin]
     
         val joinInfo = join.analyzeCondition
     
    -    JoinUtil.isStreamStreamJoin(
    -      joinInfo.getRemaining(join.getCluster.getRexBuilder),
    -      join.getRowType)
    +    try {
    +      val leftRowSchema = new RowSchema(join.getLeft.getRowType)
    +
    +      val result =
    +        WindowJoinUtil.analyzeTimeBoundary(
    +          joinInfo.getRemaining(join.getCluster.getRexBuilder),
    +          leftRowSchema.logicalType.getFieldCount,
    +          leftRowSchema.physicalType.getFieldCount,
    +          join.getRowType,
    +          join.getCluster.getRexBuilder,
    +          TableConfig.DEFAULT)
    +      timeType = result._1
    +      leftLowerBoundary = result._2
    +      leftUpperBoundary = result._3
    +      remainCondition = result._4
    +      true
    --- End diff --
    
    Hi @fhueske , I think a good reason to put the check in `translateToPlan` is for throwing
a proper exception message, such as `Left join is not supported...`. If we put the check in
rule's matches or throw the exception in rule's matches, it will throw a obscure message with
the logical plan.
    
    IMO, a meaningful message is important for users. 


> Support proctime inner equi-join between two streams in the SQL API
> -------------------------------------------------------------------
>
>                 Key: FLINK-6232
>                 URL: https://issues.apache.org/jira/browse/FLINK-6232
>             Project: Flink
>          Issue Type: Sub-task
>          Components: Table API & SQL
>            Reporter: hongyuhong
>            Assignee: hongyuhong
>
> The goal of this issue is to add support for inner equi-join on proc time streams to
the SQL interface.
> Queries similar to the following should be supported:
> {code}
> SELECT o.proctime, o.productId, o.orderId, s.proctime AS shipTime 
> FROM Orders AS o 
> JOIN Shipments AS s 
> ON o.orderId = s.orderId 
> AND o.proctime BETWEEN s.proctime AND s.proctime + INTERVAL '1' HOUR;
> {code}
> The following restrictions should initially apply:
> * The join hint only support inner join
> * The ON clause should include equi-join condition
> * The time-condition {{o.proctime BETWEEN s.proctime AND s.proctime + INTERVAL '1' HOUR}}
only can use proctime that is a system attribute, the time condition only support bounded
time range like {{o.proctime BETWEEN s.proctime - INTERVAL '1' HOUR AND s.proctime + INTERVAL
'1' HOUR}}, not support unbounded like {{o.proctime > s.protime}},  and  should include
both two stream's proctime attribute, {{o.proctime between proctime() and proctime() + 1}}
should also not be supported.
> This issue includes:
> * Design of the DataStream operator to deal with stream join
> * Translation from Calcite's RelNode representation (LogicalJoin). 



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message