SEZ9 commented on PR #11206: URL: https://github.com/apache/seatunnel/pull/11206#issuecomment-6207394196
Thanks for the detailed status on `9007896511517e4b874142cfe704102861bbc444`. Going through the points: **F3** — Good, if `scan.newly-added-table.enabled` already defaults to `false` on this head and the MySQL-CDC doc states that explicitly, then existing wildcard jobs keep their current restore behavior and I'm fine without an upgrade note for this option. I'll re-check the option declaration and the doc row on the next push and mark F3 resolved then. **F1 / F8** — The test plan sounds right: backtick and `;` cases in `MySqlSchemaTest`, plus a deserializer-level test asserting that a name failing validation produces no `CreateTableEvent` and never reaches the sink. Please also make the WARN-and-skip behavior for all-numeric names explicit in the doc alongside the identifier shape, so operators understand why such a table is not picked up rather than seeing it as a silent omission. **F2, F4, F5, F6, F7** — Understood these are still pending. For the next push I'm specifically looking for: - F2: Javadoc on both `supportsNewlyCreatedTable` declarations stating that a runtime `CreateTableEvent` is never silently dropped and that the job fails when the sink has not opted in — and the dispatch code actually enforcing that. - F4: strict routing/target-scope validation so a source-side CREATE TABLE can only produce DDL for the intended sink target. - F5: a defined retry/idempotency behavior when runtime writer creation throws `IOException`, so a transient failure cannot orphan the new table while its rows keep flowing. - F6: the restore test with an old-format `MultiTableState`. - F7: the explicit `exactly_once = false` (and `generate_sink_sql = true`) constraints in both the en and zh JDBC sink docs. **CI** — Noted that the two red groups on this head are reported as unrelated to this change. Please merge current `dev` with the next push as you proposed and share the outcome; if `all-connectors-it-2` or `engine-v2-it` still fail after that, let's look at whether there is any interaction before concluding they are pre-existing. Ping me once the push is up and I'll do the full pass on the remaining items. <!-- streview-comment:1791 --> -- 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]
