make backToBackMap of DatasetBenchmark fair

inouehrs · inouehrs · commit ca21e673916e · 2016-05-31T16:17:07.000+09:00
In the original implementation, DataFrame version processes  less work
than RDD and Dataset versions.
diff --git a/sql/core/src/test/scala/org/apache/spark/sql/DatasetBenchmark.scala b/sql/core/src/test/scala/org/apache/spark/sql/DatasetBenchmark.scala
@@ -53,7 +53,7 @@ object DatasetBenchmark {
       var res = df
       var i = 0
       while (i < numChains) {
-        res = res.select($"l" + 1 as "l")
+        res = res.select($"l" + 1 as "l", $"s")
         i += 1
       }
       res.queryExecution.toRdd.foreach(_ => Unit)

Original file line number	Diff line number	Diff line change
`@@ -53,7 +53,7 @@ object DatasetBenchmark {`
`53`	`53`	`var res = df`
`54`	`54`	`var i = 0`
`55`	`55`	`while (i < numChains) {`
`56`		`- res = res.select($"l" + 1 as "l")`
	`56`	`+ res = res.select($"l" + 1 as "l", $"s")`
`57`	`57`	`i += 1`
`58`	`58`	`}`
`59`	`59`	`res.queryExecution.toRdd.foreach(_ => Unit)`