EnricoMi commented on code in PR #1640:
URL:
https://github.com/apache/incubator-uniffle/pull/1640#discussion_r1560676747
##########
client-spark/spark3/src/main/java/org/apache/spark/shuffle/reader/RssShuffleReader.java:
##########
@@ -123,36 +124,51 @@ public RssShuffleReader(
public Iterator<Product2<K, C>> read() {
LOG.info("Shuffle read started:" + getReadInfo());
- Iterator<Product2<K, C>> aggrIter = null;
- Iterator<Product2<K, C>> resultIter = null;
+ Iterator<Product2<K, C>> resultIter;
MultiPartitionIterator rssShuffleDataIterator = new
MultiPartitionIterator<K, C>();
- if (shuffleDependency.aggregator().isDefined()) {
- if (shuffleDependency.mapSideCombine()) {
- aggrIter =
- shuffleDependency
- .aggregator()
- .get()
- .combineCombinersByKey(rssShuffleDataIterator, context);
- } else {
- aggrIter =
- shuffleDependency
- .aggregator()
- .get()
- .combineValuesByKey(rssShuffleDataIterator, context);
- }
- } else {
- aggrIter = rssShuffleDataIterator;
- }
-
if (shuffleDependency.keyOrdering().isDefined()) {
// Create an ExternalSorter to sort the data
- ExternalSorter<K, C, C> sorter =
- new ExternalSorter<>(
- context, Option.empty(), Option.empty(),
shuffleDependency.keyOrdering(), serializer);
+ ExternalSorter<K, Object, C> sorter;
+ if (shuffleDependency.aggregator().isDefined()) {
+ if (shuffleDependency.mapSideCombine()) {
+ Aggregator<K, C, C> aggregator =
+ new Aggregator<>(
+ x -> x,
+ shuffleDependency.aggregator().get().mergeCombiners(),
+ shuffleDependency.aggregator().get().mergeCombiners());
+ sorter =
+ (ExternalSorter<K, Object, C>)
+ new ExternalSorter<>(
+ context,
+ Option.apply(aggregator),
+ Option.empty(),
+ shuffleDependency.keyOrdering(),
+ serializer);
+ } else {
+ Aggregator<K, Object, C> aggregator =
+ (Aggregator<K, Object, C>) shuffleDependency.aggregator().get();
+ sorter =
+ (ExternalSorter<K, Object, C>)
+ new ExternalSorter<>(
+ context,
+ Option.apply(aggregator),
+ Option.empty(),
+ shuffleDependency.keyOrdering(),
+ serializer);
+ }
+ } else {
+ sorter =
+ new ExternalSorter<>(
+ context,
+ Option.empty(),
+ Option.empty(),
+ shuffleDependency.keyOrdering(),
+ serializer);
+ }
Review Comment:
You could simplify this (this is only sketched, not tested):
```suggestion
Option<Aggregator<K, C, C>> aggregator = Option.empty();
if (shuffleDependency.aggregator().isDefined()) {
if (shuffleDependency.mapSideCombine()) {
aggregator =
new Aggregator<>(
x -> x,
shuffleDependency.aggregator().get().mergeCombiners(),
shuffleDependency.aggregator().get().mergeCombiners());
} else {
Aggregator<K, Object, C> aggregator =
(Aggregator<K, Object, C>)
shuffleDependency.aggregator().get();
}
}
ExternalSorter<K, Object, C> sorter =
new ExternalSorter<>(
context,
aggregator,
Option.empty(),
shuffleDependency.keyOrdering(),
serializer);
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]