Github user davies commented on a diff in the pull request:
https://github.com/apache/spark/pull/11274#discussion_r55293764
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/BroadcastHashJoin.scala
---
@@ -279,17 +277,21 @@ case class BroadcastHashJoin(
s"final boolean $conditionPassed = true;"
}
+ val resultVars = buildSide match {
+ case BuildLeft => buildVars ++ input
+ case BuildRight => input ++ buildVars
+ }
if (broadcastRelation.value.isInstanceOf[UniqueHashedRelation]) {
s"""
|// generate join key for stream side
|${keyEv.code}
|// find matches from HashedRelation
|UnsafeRow $matched = $anyNull ? null:
(UnsafeRow)$relationTerm.getValue(${keyEv.value});
- |${buildVars.map(_.code).mkString("\n")}
|${checkCondition.trim}
|if (!$conditionPassed) {
- | // reset to null
- | ${buildVars.map(v => s"${v.isNull} = true;").mkString("\n")}
+ | $matched = null;
+ | // reset the variables those are already evaluated.
--- End diff --
If we find a row on build side, but the condition do not pass, the
variables for build side still have the value (from build side), but the result
should be empty (outer join).
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
