Github user aljoscha commented on a diff in the pull request:
https://github.com/apache/incubator-flink/pull/263#discussion_r21690839
--- Diff:
flink-scala/src/main/java/org/apache/flink/api/scala/operators/ScalaAggregateOperator.java
---
@@ -164,18 +160,8 @@ public ScalaAggregateOperator(Grouping<IN> input,
Aggregations function, int fie
}
genName.setLength(genName.length()-1);
- TypeSerializer<IN> serializer =
getInputType().createSerializer();
- TypeSerializerFactory<IN> serializerFactory;
- if (serializer.isStateful()) {
- serializerFactory = new
RuntimeStatefulSerializerFactory<IN>(
- serializer,
getInputType().getTypeClass());
- } else {
- serializerFactory = new
RuntimeStatelessSerializerFactory<IN>(
- serializer,
getInputType().getTypeClass());
- }
-
@SuppressWarnings("rawtypes")
- RichGroupReduceFunction<IN, IN> function = new
AggregatingUdf(serializerFactory, aggFunctions, fields);
+ RichGroupReduceFunction<IN, IN> function = new
AggregatingUdf((TupleSerializerBase) getInputType().createSerializer(),
aggFunctions, fields);
--- End diff --
We could maybe. If we enhanced TypeSerializer to have a method
createInstance(...) that can create an instance with the given field values.
This is only possible for a small subset of the types that we support: The Java
Tuple Types, since we control their constructors. And Scala Case Classes, since
we assume that the fields are exactly the constructor arguments, as is
customary for Case Classes.
We would then need to add createInstance(...) methods to all other
TypeSerializers that throw an Exception telling the user that instance creation
with field values is not supported.
---
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.
---
