DanielLeens commented on issue #11739: URL: https://github.com/apache/seatunnel/issues/11739#issuecomment-5239946418
Thanks for opening this follow-up audit. I re-checked current `dev`, and the issue is real as a connector-by-connector bounded-read gap: - MySQL CDC now exposes `stop.mode` values `never`, `latest`, and `specific`. - Postgres, Oracle, SQL Server, and MongoDB still expose only `never` in their public `stop.mode` options today. So this is not just a documentation mismatch. The MySQL work has moved the capability forward, but the other `FetchTask`-based connectors still do not expose an end-to-end bounded-read contract yet. I would keep this issue as the audit / umbrella rather than turn it into one giant implementation PR. The next highest-value step is the matrix you outlined: - per connector, confirm the stop-offset ordering semantics are strong enough for comparison; - confirm where the streaming task can actually observe and enforce the stop boundary; - define the E2E proof that the job reaches `FINISHED` rather than just stopping emission while the reader keeps running. Once that is written down, follow-up PRs can stay one connector at a time instead of mixing Oracle / PostgreSQL / SQL Server / MongoDB semantics into one review. -- 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]
