[SPARK-13760][SQL] Fix BigDecimal constructor for FloatType

sameeragarwal · yhuai · commit 926e9c45a21c · 2016-03-09T18:16:29.000-08:00
## What changes were proposed in this pull request? A very minor change for using `BigDecimal.decimal(f: Float)` instead of `BigDecimal(f: float)`. The latter is deprecated and can result in inconsistencies due to an implicit conversion to `Double`. ## How was this patch tested? N/A cc yhuai Author: Sameer Agarwal <sameer@databricks.com> Closes #11597 from sameeragarwal/bigdecimal.
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/mathExpressions.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/mathExpressions.scala
@@ -748,7 +748,7 @@ case class Round(child: Expression, scale: Expression)
         if (f.isNaN || f.isInfinite) {
           f
         } else {
-          BigDecimal(f).setScale(_scale, HALF_UP).toFloat
+          BigDecimal.decimal(f).setScale(_scale, HALF_UP).toFloat
         }
       case DoubleType =>
         val d = input1.asInstanceOf[Double]

Original file line number	Diff line number	Diff line change
`@@ -748,7 +748,7 @@ case class Round(child: Expression, scale: Expression)`
`748`	`748`	`if (f.isNaN \|\| f.isInfinite) {`
`749`	`749`	`f`
`750`	`750`	`} else {`
`751`		`- BigDecimal(f).setScale(_scale, HALF_UP).toFloat`
	`751`	`+ BigDecimal.decimal(f).setScale(_scale, HALF_UP).toFloat`
`752`	`752`	`}`
`753`	`753`	`case DoubleType =>`
`754`	`754`	`val d = input1.asInstanceOf[Double]`