linliu-code commented on PR #19915: URL: https://github.com/apache/hudi/pull/19915#issuecomment-5638997710
Closing: the reachable scenario turned out to be a corner setup, and the end-to-end evidence did not materialise. What was established: - The `shouldCombine &&` gate here is the same conflation fixed for the Spark SQL path, and `TestHoodieStreamerUtils#testInsertWithCustomPayloadCarriesTheOrderingValue` does pin it at record construction, proved by reverting. - But reaching the actual merge failure needs `--op INSERT` plus a payload class outside `DEPRECATED_PAYLOADS`, and a DeltaStreamer functional test built for exactly that scenario passes with **and** without the fix. So the path from "record carries the wrong ordering value" to "a merge compares it against a Long" is not established on the Streamer side. - `UPSERT` computes the value already; `INSERT` and `BULK_INSERT` with a built-in payload produce a `HoodieAvroIndexedRecord`, which derives it correctly. Not worth carrying an unproven fix for a configuration we have not shown anyone runs. The Spark SQL path fix, which has a demonstrated failure, is onehouseinc/hudi-internal#2412 and apache/hudi#19908. Reopen if a real case turns up. -- 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]
