cloud-fan commented on a change in pull request #26400: [SPARK-29759][SQL]
local/coalesced shuffle reader shouldn't strip itself when canonicalization
URL: https://github.com/apache/spark/pull/26400#discussion_r342610113
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/adaptive/OptimizeLocalShuffleReader.scala
##########
@@ -103,25 +103,11 @@ case class LocalShuffleReaderExec(child: QueryStageExec)
extends UnaryExecNode {
override def output: Seq[Attribute] = child.output
- override def doCanonicalize(): SparkPlan = child.canonicalized
-
- override def outputPartitioning: Partitioning = {
-
- def tryReserveChildPartitioning(stage: ShuffleQueryStageExec):
Partitioning = {
- val initialPartitioning = stage.plan.child.outputPartitioning
- if (initialPartitioning.isInstanceOf[UnknownPartitioning]) {
- UnknownPartitioning(stage.plan.shuffleDependency.rdd.partitions.length)
Review comment:
I think this is over complicated. We never leverage `UnknownPartitioning` in
Spark SQL. It's better to make the logic simpler and just return the output
partitioning of the underlying plan.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
