DanielLeens commented on PR #11746: URL: https://github.com/apache/seatunnel/pull/11746#issuecomment-5611962561
Thanks for closing the loop on the dialect verification matrix, @li3zhi4 — good to have MySQL 7/7 and Oracle 3/3 confirmed green on `c8c4c3069` (and glad the earlier MySQL failure was just an image-pull timeout, not a real regression). To be clear on scope though: this verification matrix doesn't touch **Issue 1** from my review above (`buildCompositeCondition`'s middle/last-split branches missing the `IS NOT NULL` guard, causing a duplicate read for NULL-secondary-key rows whose leading column doesn't tie to a split boundary). That's a code-level gap, not a test-flakiness one — `testCompositeKeyWithNullPrimaryKeyComponent`'s current fixture (all rows sharing one of 3 `order_id` values) structurally can't exercise the failure path I traced, so a green run on this head doesn't clear it. The fix is still the one-line-per-branch `AND buildNotNullKeyCondition(columnNames)` addition to the `isLastSplit` and middle-split branches, plus tightening the unit test to check the actual guard clause and giving the SQLite E2E fixture enough distinct leading-column values to land a NULL row strictly inside a middle split. Once that lands I'll do a fast, scoped re-check (this is a localized fix, doesn't need a full re-review) rather than another full round. -- 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]
