viirya commented on code in PR #774:
URL: https://github.com/apache/datafusion-comet/pull/774#discussion_r1702941755
##########
spark/src/main/scala/org/apache/spark/sql/comet/CometWindowExec.scala:
##########
@@ -65,71 +65,27 @@ case class CometWindowExec(
sparkContext,
"number of partitions")) ++ readMetrics ++ writeMetrics
- override def supportsColumnar: Boolean = true
-
- protected override def doExecuteColumnar(): RDD[ColumnarBatch] = {
- val childRDD = child.executeColumnar()
-
- childRDD.mapPartitionsInternal { iter =>
- CometExec.getCometIterator(
- Seq(iter),
- getNativePlan(output, windowExpression, partitionSpec, orderSpec,
child).get)
- }
- }
-
override def outputOrdering: Seq[SortOrder] = child.outputOrdering
override def outputPartitioning: Partitioning = child.outputPartitioning
protected def withNewChildInternal(newChild: SparkPlan): SparkPlan =
this.copy(child = newChild)
-}
-
-object CometWindowExec {
- def getNativePlan(
- outputAttributes: Seq[Attribute],
- windowExpression: Seq[NamedExpression],
- partitionSpec: Seq[Expression],
- orderSpec: Seq[SortOrder],
- child: SparkPlan): Option[Operator] = {
-
- val orderSpecs = orderSpec.map(exprToProto(_, child.output))
- val partitionSpecs = partitionSpec.map(exprToProto(_, child.output))
- val scanBuilder = OperatorOuterClass.Scan.newBuilder()
- val scanOpBuilder = OperatorOuterClass.Operator.newBuilder()
-
- val scanTypes = outputAttributes.flatten { attr =>
- serializeDataType(attr.dataType)
+ override def stringArgs: Iterator[Any] =
+ Iterator(output, windowExpression, partitionSpec, orderSpec, child)
+
+ override def equals(obj: Any): Boolean = {
+ obj match {
+ case other: CometWindowExec =>
+ this.windowExpression == other.windowExpression && this.child ==
other.child &&
+ this.partitionSpec == other.partitionSpec && this.orderSpec ==
other.orderSpec &&
+ this.serializedPlanOpt == other.serializedPlanOpt
+ case _ =>
+ false
}
Review Comment:
Besides, native window operator requires correct partitioning and ordering
information from its child operator. The current native scan operator doesn't
have these information. That also causes DataFusion Window operator cannot
produce correct ordering partition by info.
--
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]
