SEZ9 commented on PR #12608: URL: https://github.com/apache/seatunnel/pull/12608#issuecomment-6029386424
Thanks @CodeWithPravinMaske, this covers what I asked for. 1. **Restore E2E** – This is exactly the scenario behind F1/F2: the `ALTER TABLE uk_added_null ADD UNIQUE KEY` in the binlog phase, then the schema coming back from the table history in the savepoint. Asserting NULL (and no `0`) after the restore for both tables is the right check, and sharing the binlog insert step with `testMysqlCdcNullInNullableUniqueKeyWithoutPrimaryKey` keeps it tidy. Zeta-only is fine, consistent with the other restore tests. 2. **Override Javadoc** – Good. Stating that it is a copy of Debezium 1.9.8, that `parsePrimaryIndexColumnNames` is the only modified method, and that the change must be re-applied or the copy dropped on a Debezium upgrade is what the next person touching that file needs. The method-level explanation makes the intent clear. 3. **F4 assertion** – Going through `CatalogTableUtils.mergeCatalogTableConfig(catalogTable, config)` rather than hand-building the table exercises the real `table-names-config` path, and checking both that the Debezium column stays optional and that a NULL `code` converts to NULL closes F4. Good that this test and the other nullability tests fail with the original Debezium behaviour, so they actually guard the fix. Two small things before I approve: - Since the F4 test shows `table-names-config.primaryKeys` on a nullable column now keeps the column optional, please make sure the MySQL-CDC docs wording and the release/incompatible-changes note (F3/F6) reflect that behaviour rather than the earlier "still coerces NULL to the type default" description. If you already updated them, just point me to it. - You mentioned the restore E2E passes locally; please confirm it is also green in CI, since it depends on savepoint timing. Once those are confirmed I'm happy with this. <!-- streview-comment:1562 --> -- 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]
