danny0405 commented on code in PR #13115:
URL: https://github.com/apache/hudi/pull/13115#discussion_r2036316223
##########
hudi-common/src/main/java/org/apache/hudi/common/table/read/FileGroupRecordBuffer.java:
##########
@@ -275,43 +274,47 @@ protected Option<Pair<Option<T>, Map<String, Object>>>
doProcessNextDataRecord(T
} else {
switch (recordMergeMode) {
case COMMIT_TIME_ORDERING:
- return Option.empty();
+ return Option.of(Pair.of(Option.ofNullable(record), metadata));
case EVENT_TIME_ORDERING:
- Comparable existingOrderingValue = readerContext.getOrderingValue(
- existingRecordMetadataPair.getLeft(),
existingRecordMetadataPair.getRight(),
- readerSchema, orderingFieldName);
- if
(isDeleteRecordWithNaturalOrder(existingRecordMetadataPair.getLeft(),
existingOrderingValue)) {
- return Option.empty();
- }
- Comparable incomingOrderingValue = readerContext.getOrderingValue(
- Option.of(record), metadata, readerSchema, orderingFieldName);
- if (incomingOrderingValue.compareTo(existingOrderingValue) > 0) {
+ if (shouldKeepNewerRecord(existingRecordMetadataPair.getLeft(),
existingRecordMetadataPair.getRight(), Option.ofNullable(record), metadata)) {
return Option.of(Pair.of(Option.of(record), metadata));
}
return Option.empty();
case CUSTOM:
default:
// Merge and store the combined record
- // Note that the incoming `record` is from an older commit, so it
should be put as
- // the `older` in the merge API
if (payloadClass.isPresent()) {
+ if (existingRecordMetadataPair.getLeft().isEmpty()
Review Comment:
I think we can do it for event time merging, the records coming from `DELETE
FROM` statements should be padded with the ordering vals so that we can do
event time merging with it. This should be a quite cool feature and we can
remove the hacky constant 0 and the same type class check.
--
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]
