DanielLeens commented on issue #10203: URL: https://github.com/apache/seatunnel/issues/10203#issuecomment-5228704446
Thanks for writing the design direction down so concretely. I re-checked the current `dev` state before replying here: I still do not see a linked PR for this issue, and I also do not see an existing generic `discover-new-tables` style option/path already exposed in the MySQL CDC source today. So this is still open design/implementation scope rather than something already covered elsewhere. Your proposed direction looks reasonable for the first implementation slice: 1. keep the admission / ownership logic in the CDC base layer rather than making the final API shape MySQL-only; 2. keep the enumerator as the single authority for admitting new tables and persisting that admitted-table / ownership state in checkpoints; 3. keep the first version default-off behind an explicit config; 4. and keep the delivery staged so the first PR stays on the safer boundary. For that first PR, I would still recommend narrowing the scope to startup / restore admission of newly matching tables plus the checkpoint-state pieces needed to keep recovery deterministic. Live binlog-time discovery of brand-new tables while the job is already running is the higher-risk part and is better handled as a follow-up once the state / ownership contract is proven in review. If you are still planning to take it, the next best step is to open a Draft PR with: - the config contract you want to introduce; - the checkpoint / admitted-table state shape; - and tests covering matching / non-matching tables, recovery, and duplicate admission. That will give maintainers something concrete to review without letting the runtime-discovery slice expand too early. -- 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]
