SEZ9 commented on issue #12821: URL: https://github.com/apache/seatunnel/issues/12821#issuecomment-6228800935
Thanks for the follow-up, and thanks for putting up #12840. The analysis matches the issue description: `ReduceBufferedBatchStatementExecutor` keys its `LinkedHashMap` buffer by the extracted `primary_key` row, so rows whose key contains NULL compare equal and collapse into a single entry before any SQL runs. Keeping every incomplete-key row in the buffer while retaining last-write-wins reduction for complete keys is the right policy. Could you confirm that #12840 includes the regression coverage requested above: a MergeTree table with `allow_nullable_key = 1`, several rows whose configured key is NULL, both the insert path and the delete/update path, and a composite key with one NULL component? It is also good that the PR calls out the separate NULL-key SQL-delete limitation; please make sure it is clearly documented so users are not surprised by it. I'll take a look at the PR. <!-- streview-comment:1816 --> -- 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]
