SEZ9 commented on PR #12079: URL: https://github.com/apache/seatunnel/pull/12079#issuecomment-5862568533
Thanks for restating these directly — that helps. **F1** — The three tests you listed (`ConfigParserUtilTest.java:356`, `MultipleTableJobConfigParserTest.java:773`, `SeaTunnelConfValidateCommandTest.java:291`) target exactly the scenario I raised: two transforms defaulting to the same output ID feeding a multi-input dependent. I'll verify at `3df75448c` that they assert on scheduling order rather than only the final result before marking this resolved. **F2 / F4** — The `emptyInputFallback = transform.inputIds.isEmpty() || transform.inputOmitted` condition at `TransformDependencyScheduler.java:140` (with `inputOmitted` at `:236`) is the right shape for removing the transform-count gating. I'll confirm the code and the tests at `MultipleTableJobConfigParserTest.java:728` / `:733` and `SeaTunnelConfValidateCommandTest.java:245` / `:273` against the head before closing these out. **F3 onward** — On my side the F3 paragraph still ends mid-sentence, right after "`ConfigParserUtil.getInputIds` (`ConfigParserUtil.java:269`) is the only place `plugin_input` is resolved from `ReadonlyConfig`; `S", and I don't see the F5/F7/F8 recap or the F6 answer after it. Whatever the cause, rather than guess, could you post them again as a short separate comment per item? 1. **F3** — the rest of the "one path" argument: which call site the dry-run scheduler now uses so `getTransformInputIds` and `getInputIds` can't diverge. 2. **F6** — with the fallback only triggering on an empty/omitted `plugin_input`, does a single-transform self-cycle (explicit `plugin_input` equal to its own output) now hard-fail in dry-run the same way the runtime parser does? A test name is enough. 3. **F5** — is `TransformDependencyScheduler` the shared scheduler used by both `DryRunConnectValidator` and `MultipleTableJobConfigParser`, or does each still carry its own copy? 4. **F7** — where the docs/upgrade-note change for the new fail-fast behaviour landed, or confirmation it's intentionally deferred. 5. **F8** — whether the redundant static helpers and the unused `getOutputId()` accessor were removed. Once I have those I'll do a final pass over the head. <!-- streview-comment:1373 --> -- 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]
