Github user xwu0226 commented on a diff in the pull request:
https://github.com/apache/spark/pull/14842#discussion_r78270996
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/SortOrder.scala
---
@@ -71,12 +95,33 @@ case class SortPrefix(child: SortOrder) extends
UnaryExpression {
val nullValue = child.child.dataType match {
case BooleanType | DateType | TimestampType | _: IntegralType =>
- Long.MinValue
+ if ((child.isAscending && child.nullOrdering == NullFirst) ||
+ (!child.isAscending && child.nullOrdering == NullLast)) {
+ Long.MinValue
+ } else {
+ Long.MaxValue
+ }
case dt: DecimalType if dt.precision - dt.scale <=
Decimal.MAX_LONG_DIGITS =>
- Long.MinValue
+ if ((child.isAscending && child.nullOrdering == NullFirst) ||
+ (!child.isAscending && child.nullOrdering == NullLast)) {
+ Long.MinValue
+ } else {
+ Long.MaxValue
+ }
case _: DecimalType =>
- DoublePrefixComparator.computePrefix(Double.NegativeInfinity)
- case _ => 0L
+ if ((child.isAscending && child.nullOrdering == NullFirst) ||
+ (!child.isAscending && child.nullOrdering == NullLast)) {
+ DoublePrefixComparator.computePrefix(Double.NegativeInfinity)
+ } else {
+ DoublePrefixComparator.computePrefix(Double.NaN)
--- End diff --
I tried the PositiveInfinity before but the randomized test sometimes fails
me. Then, I view the actual sorted array and found that `NaN` is always
considered bigger than `PositiveIntinity`. So I changed to use `NaN` here.
---
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]
