Allow zero partition count in HashPartitioner

There are valid cases where 0-size HashPartitioners are created (such as running groupByKey on an empty RDD). As long as they don't call getPartition there is nothing wrong with this. getPartition will try to divide by zero when it is called in this case, so there is no risk of silent mistakes. For negative partition counts getPartition would return bogus results, so the assertion against that remains.

Author: darabos

core/src/main/scala/org/apache/spark/Partitioner.scala

@@ -76,7 +76,7 @@ object Partitioner {
  * produce an unexpected or incorrect result.
  */
 class HashPartitioner(partitions: Int) extends Partitioner {
-  require(partitions > 0, s"Number of partitions ($partitions) must be positive.")
+  require(partitions >= 0, s"Number of partitions ($partitions) cannot be negative.")
 
   def numPartitions: Int = partitions