SEZ9 commented on issue #11729: URL: https://github.com/apache/seatunnel/issues/11729#issuecomment-5902933518
Thanks @SiddhantShedge45 and @zhang-arvin for offering to help here. As @siwen-yu already pointed out, there is an existing PR (#11730) for this bug that is waiting on maintainer review, so we won't reassign the issue at this point. The expected fix is straightforward: when every column of the target table is part of the key, the generated `MERGE` must not emit an empty `WHEN MATCHED THEN UPDATE SET` clause — the matched branch should be dropped entirely so the statement stays valid and unmatched rows are still inserted. If you'd like to contribute, the most useful thing right now is to review that PR and independently verify the fix against the affected `MERGE`-based dialects (DM, Oracle, SAP HANA, SQL Server, Vertica, Xugu) with an all-key table. Reproduction results or gaps you find in the dialect coverage would be very welcome as review comments there. @siwen-yu, a couple of concrete asks so we can move the PR forward: - Please confirm that the change covers all six dialects listed above, not only the one you originally hit. - Please make sure there is a regression test for the all-key-columns case that asserts the matched branch is omitted (and that the insert path still works). We'll keep this issue open until the PR lands. If it stalls, we can revisit assignment then. <!-- streview-comment:1417 --> -- 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]
