DanielLeens commented on PR #11912: URL: https://github.com/apache/seatunnel/pull/11912#issuecomment-6592033555
Thanks for the status update, @SEZ9, and for keeping the threads organized. I re-checked your list against the unchanged head (`d3a8a348ddb0`, still no new commit since 2026-10-22), and I think items 1 to 5 are already addressed in the branch, so they should not block @dybyte after the rebase. They match what I verified on 2026-10-25 (and what you mapped back in your follow-up on that day): 1. Lazy applier creation: `MultiTableSink.createWriter()` / `restoreWriter()` no longer build appliers; `createSchemaChangeApplier(TablePath)` builds one only when a coordinated DDL is dispatched. 2. Deterministic resolution and keying: logical-key lookup, then `resolvePhysicalSinkTable(...)`, which throws an `IOException` on ambiguous physical matches; the applier is created for the resolved physical path. 3. Identifier safety: documented in the `SchemaChangeApplier.apply()` javadoc. 4. Writer contract: `FlinkSink` fails fast when a sink declares coordinated support but its writer does not implement `SupportSchemaRefreshSinkWriter`. 5. Docs: en/zh updates in `sink-connector-development.md`, `schema-evolution.md` and the checkpoint/savepoint restore note in `incompatible-changes.md`. If you see a spot where one of these is not covered at the current head, please point me at the line and I will re-check. What is genuinely still open: - The merge conflict with `dev` (`CONFLICTING` / `DIRTY`) and the required `Build` check at `ACTION_REQUIRED`. - The P1 from @nzw921rx on `SchemaOperator` / `SchemaRefreshBarrierOperator` (a later completed checkpoint can release buffered post-DDL rows before writer refresh, with `max-concurrent-checkpoints > 1`). - One small non-blocking javadoc line on `SupportCoordinatedSchemaEvolutionSink.supportsCoordinatedSchemaEvolution()` explaining why a plain `instanceof` is not enough. @dybyte, thank you for the work so far. Could you sync the latest `dev`, resolve the conflicts (please double-check that the `MultiTableSink`, `SchemaChangeApplier` and `SupportCoordinatedSchemaEvolutionSink` changes are neither lost nor duplicated during the merge), and enable GitHub Actions in your fork so the checks can run on the new head? Then we can take another look at the P1 and the javadoc line together. -- 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]
