Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/2025#discussion_r64542323
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/api/table/plan/nodes/dataset/DataSetUnion.scala
---
@@ -69,16 +73,23 @@ class DataSetUnion(
rows + metadata.getRowCount(child)
}
- planner.getCostFactory.makeCost(rowCnt, 0, 0)
+ planner.getCostFactory.makeCost(
+ rowCnt,
+ if (all) 0 else rowCnt,
+ if (all) 0 else rowCnt)
}
override def translateToPlan(
tableEnv: BatchTableEnvironment,
expectedType: Option[TypeInformation[Any]]): DataSet[Any] = {
- val leftDataSet =
left.asInstanceOf[DataSetRel].translateToPlan(tableEnv)
- val rightDataSet =
right.asInstanceOf[DataSetRel].translateToPlan(tableEnv)
- leftDataSet.union(rightDataSet).asInstanceOf[DataSet[Any]]
+ val leftDataSet =
left.asInstanceOf[DataSetRel].translateToPlan(tableEnv, expectedType)
+ val rightDataSet =
right.asInstanceOf[DataSetRel].translateToPlan(tableEnv, expectedType)
--- End diff --
Actually, this fix should be extended. If `expectedType == None`, we must
ensure that both inputs provide the same type, i.e., request the type of the
left input on the right input (or vice versa).
---
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.
---
