JingsongLi commented on PR #9370: URL: https://github.com/apache/paimon/pull/9370#issuecomment-5748031732
Requirement fit: SUPPORTED. Implementation: CLEAN in the reviewed follow-up. Per-partition bucket counts have clear end-to-end value for independently rescaling hot/cold partitions without rewriting the whole table. The two previously identified unsafe paths are now closed: Spark rejects both V1/command and V2 writes when the feature is enabled instead of hashing with the table-level bucket count, and `PartitionEntry.merge` fails when equal creation times carry conflicting positive bucket layouts instead of silently choosing the larger count. The Flink writer/coordinator, restore, overwrite and rescale paths have dedicated tests, and the current Java/Flink/Spark/E2E CI matrix is green. I did not find a new evidence-backed blocker in the author follow-up. The Spark restriction should remain prominently documented until Spark gains partition-aware routing. -- 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]
