JunRuiLee commented on code in PR #7832:
URL: https://github.com/apache/paimon/pull/7832#discussion_r3258545859
##########
paimon-core/src/main/java/org/apache/paimon/mergetree/compact/PartialUpdateMergeFunction.java:
##########
@@ -358,7 +362,8 @@ public KeyValue getResult() {
}
RowKind rowKind = currentDeleteRow || !meetInsert ? RowKind.DELETE :
RowKind.INSERT;
- return reused.replace(currentKey, latestSequenceNumber, rowKind, row);
+ return reused.replace(currentKey, latestSequenceNumber, rowKind, row)
Review Comment:
I don’t think this case is reachable in normal compaction.
Partial-update without sequence group relies on the merge-tree invariant
that compaction picks a run-based prefix, not arbitrary files; so it cannot
compact snapshot 1 and snapshot 3 while leaving snapshot 2 for the same key
outside.
If such a non-contiguous compaction were possible, regular row-level
sequence number ordering would already have the same correctness issue, not
only snapshot ordering.
--
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]
