[
https://issues.apache.org/jira/browse/FLINK-4231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15426663#comment-15426663
]
ASF GitHub Bot commented on FLINK-4231:
---------------------------------------
Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/2272#discussion_r75332918
--- Diff:
flink-core/src/main/java/org/apache/flink/api/common/operators/base/ReduceOperatorBase.java
---
@@ -202,7 +203,9 @@ public void setCustomPartitioner(Partitioner<?>
customPartitioner) {
if (inputColumns.length > 0) {
boolean[] inputOrderings = new
boolean[inputColumns.length];
- TypeComparator<T> inputComparator = ((CompositeType<T>)
inputType).createComparator(inputColumns, inputOrderings, 0, executionConfig);
+ TypeComparator<T> inputComparator = inputType
instanceof BasicTypeInfo
--- End diff --
Check for `AtomicType` instead of `BasicTypeInfo`?
> Switch DistinctOperator from GroupReduce to Reduce
> --------------------------------------------------
>
> Key: FLINK-4231
> URL: https://issues.apache.org/jira/browse/FLINK-4231
> Project: Flink
> Issue Type: Improvement
> Components: Java API
> Affects Versions: 1.1.0
> Reporter: Greg Hogan
> Assignee: Greg Hogan
>
> As discussed in FLINK-3279, rewriting {{DistinctOperator}} to a
> {{ReduceFunction}} rather than the current {{GroupReduceFunction}} allows the
> user to set the {{CombineHint}} and choose a hash-based, sort-based, or no
> combiner.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
