leanken commented on a change in pull request #29104:
URL: https://github.com/apache/spark/pull/29104#discussion_r460676977
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/joins/HashedRelation.scala
##########
@@ -896,22 +967,29 @@ private[joins] object LongHashedRelation {
// Create a mapping of key -> rows
var numFields = 0
+ val isOriginalInputEmpty: Boolean = !input.hasNext
+ var allNullColumnKeyExistsInOriginalInput: Boolean = false
while (input.hasNext) {
val unsafeRow = input.next().asInstanceOf[UnsafeRow]
numFields = unsafeRow.numFields()
val rowKey = keyGenerator(unsafeRow)
if (!rowKey.isNullAt(0)) {
val key = rowKey.getLong(0)
map.append(key, unsafeRow)
+ } else if (!allNullColumnKeyExistsInOriginalInput) {
+ allNullColumnKeyExistsInOriginalInput = true
}
}
map.optimize()
new LongHashedRelation(numFields, map)
+ .setOriginalInputEmtpy(isOriginalInputEmpty)
+
.setAllNullColumnKeyExistsInOriginalInput(allNullColumnKeyExistsInOriginalInput)
+ .asInstanceOf[LongHashedRelation]
}
}
/** The HashedRelationBroadcastMode requires that rows are broadcasted as a
HashedRelation. */
-case class HashedRelationBroadcastMode(key: Seq[Expression])
+case class HashedRelationBroadcastMode(key: Seq[Expression], isNullAware:
Boolean = false)
Review comment:
let me try update to adopt EmptyHashedRelation and get rid of original
input
----------------------------------------------------------------
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
