Github user marmbrus commented on a diff in the pull request:
https://github.com/apache/spark/pull/1238#discussion_r14677118
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/SparkStrategies.scala ---
@@ -61,8 +61,6 @@ private[sql] abstract class SparkStrategies extends
QueryPlanner[SparkPlan] {
condition.map(Filter(_,
broadcastHashJoin)).getOrElse(broadcastHashJoin) :: Nil
}
- def broadcastTables: Seq[String] =
sqlContext.joinBroadcastTables.split(",").toBuffer
-
def apply(plan: LogicalPlan): Seq[SparkPlan] = plan match {
--- End diff --
I don't see a need for two config options here. -1 as the threshold works
as a perfectly reasonable way to turn off auto-broadcasting.
I agree that it would be a good idea to set the build side for hash joins
to the relation with the smaller estimated size. If its not too hard we should
just do that in this PR too.
We should add comments to the scala doc that explain the estimation magic
that is going on here now.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
