Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/3795#discussion_r22552583
--- Diff:
core/src/main/scala/org/apache/spark/rdd/OrderedRDDFunctions.scala ---
@@ -72,6 +72,8 @@ class OrderedRDDFunctions[K : Ordering : ClassTag,
* because it can push the sorting down into the shuffle machinery.
*/
def repartitionAndSortWithinPartitions(partitioner: Partitioner):
RDD[(K, V)] = {
+ val keyClass = implicitly[ClassTag[K]].runtimeClass
+ Partitioner.assertHashCodeIsWellBehaved(keyClass)
--- End diff --
I think the concern here is that there's other code which might call
`hashCode` on the items, so even if we use a custom partitioner we will still
get incorrect results for Java arrays. I think that this is safe as long as we
can guarantee that the sorting phase doesn't call `hashCode`.
---
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]
