Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/3641#discussion_r108863035
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/aggregate/AggregateUtil.scala
---
@@ -1208,5 +1208,40 @@ object AggregateUtil {
private def gcd(a: Long, b: Long): Long = {
if (b == 0) a else gcd(b, a % b)
}
-}
+
+ /**
+ * Create an
[[org.apache.flink.streaming.api.functions.ProcessFunction]] to evaluate final
+ * aggregate value over a window with processing time boundaries.
+ *
+ * @param namedAggregates List of calls to aggregate functions and their
output field names
+ * @param inputType Input row type
+ * @param timeBoundary time limit of the window boundary expressed in
milliseconds
+ * @param isPartitioned Flag to indicate whether the input is
partitioned or not
+ * @return [[org.apache.flink.streaming.api.functions.ProcessFunction]]
+ */
+ private[flink] def createTimeBoundedProcessingOverProcessFunction(
+ namedAggregates: Seq[CalcitePair[AggregateCall, String]],
+ inputType: RelDataType,
+ timeBoundary: Long,
+ isPartitioned: Boolean = true): ProcessFunction[Row, Row] = {
+
+ val (aggFields, aggregates) =
+ transformToAggregateFunctions(
+ namedAggregates.map(_.getKey),
+ inputType,
+ needRetraction = false)
--- End diff --
The parameter asks for AggregateFunctions that support retraction. There
are some functions (like min and max) that are less efficient if they need to
prepare for retraction. Therefore they have a retractable and a non-retractable
version. The parameter decides which version is returned.
---
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.
---
