DanielLeens commented on issue #12983: URL: https://github.com/apache/seatunnel/issues/12983#issuecomment-6397378856
Thanks for separating the Flink/Spark adapter work from #11932. I checked current `dev`: `FlinkRowCollector` and `InternalRowCollector` have no `restoreSchema` override, while #11932 explicitly limits its collector recovery claim to Zeta. That identifies a validation gap, not proof that both adapters fail at runtime. I have marked this as `help wanted` for a focused contribution; there is no assignee or adapter PR in this lane yet. A useful first PR should trace where one adapter owns its restored source schema and downstream row conversion, then run a CDC schema-change checkpoint/savepoint restore through that adapter and assert the columns, types, and values of the *first* emitted row. Include the schema-change-disabled and legacy-checkpoint cases, and show that refreshing local schema state does not replay downstream DDL. Flink and Spark may be separate PRs if their state ownership differs; please state which adapter and recovery boundary each PR actually validates. Please link any PR here and report the relevant integration-suite result on its final head. #11932 is still open with a failing Build, so its Zeta validation should stay separate from the evidence for these adapters. -- 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]
