xushiyan commented on a change in pull request #1187: [HUDI-499] Allow update
partition path with GLOBAL_BLOOM
URL: https://github.com/apache/incubator-hudi/pull/1187#discussion_r371022749
##########
File path:
hudi-client/src/main/java/org/apache/hudi/index/bloom/HoodieGlobalBloomIndex.java
##########
@@ -114,14 +117,24 @@ public HoodieGlobalBloomIndex(HoodieWriteConfig config) {
keyLocationPairRDD.mapToPair(p -> new Tuple2<>(p._1.getRecordKey(),
new Tuple2<>(p._2, p._1)));
// Here as the recordRDD might have more data than rowKeyRDD (some
rowKeys' fileId is null), so we do left outer join.
- return
incomingRowKeyRecordPairRDD.leftOuterJoin(existingRecordKeyToRecordLocationHoodieKeyMap).values().map(record
-> {
+ return
incomingRowKeyRecordPairRDD.leftOuterJoin(existingRecordKeyToRecordLocationHoodieKeyMap).values().flatMap(record
-> {
final HoodieRecord<T> hoodieRecord = record._1;
final Optional<Tuple2<HoodieRecordLocation, HoodieKey>>
recordLocationHoodieKeyPair = record._2;
if (recordLocationHoodieKeyPair.isPresent()) {
// Record key matched to file
- return getTaggedRecord(new
HoodieRecord<>(recordLocationHoodieKeyPair.get()._2, hoodieRecord.getData()),
Option.ofNullable(recordLocationHoodieKeyPair.get()._1));
+ if (config.getGlobalBloomIndexShouldUpdatePartitionPath()
Review comment:
2nd thought on this: we should treat a configuration as the dominating
factor and the partition equality check as a supporting factor. Doing it the
other way will diminish the purpose of a configuration. Hence I would keep that
order as is. Thanks for bringing thoughts to it!
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
