SEZ9 commented on PR #12824: URL: https://github.com/apache/seatunnel/pull/12824#issuecomment-6995344156
Thanks for the test migration and for documenting extension identity. The REST and CLI output tests now pin the mixed-chain behavior. The earlier running/failing CI guidance no longer applies, since attempt 2 of the current-head Build (run 44137036919) completed successfully. That clears the CI gate only. Three source-contract points are still open on this head: 1. **Migration path.** `OptionRule.Builder` has no method that accepts a `Condition` chain as the trigger, so the migration guide names constructors whose result cannot be attached to an `OptionRule` through any public builder path. Please either add that builder entry point or adjust the guide so it only describes a path that works. 2. **Null triggers.** `new ConditionRule(null, rule)` used to fail during validation and is now silently skipped, while the REST and CLI exporters throw a `NullPointerException` on the same rule. `ConditionalRequiredOptions.of(null, ...)` is skipped by the validator in the same way. Please make the validator and the exporters behave consistently for a null trigger, and add a test for it. 3. **Extension equality migration note.** `Condition.equals`/`hashCode` now include the `ConditionExtension` instance, which changes `Condition` and `OptionRule` equality for every factory that creates its extension inline. Extension identity is now documented, but this note is still recorded as open. Could you point to where the incompatible-changes entry covers it, or add it there? The latest comments do not say where the other three previous points stand, so could you confirm each one? - **Parentheses in rendered chains.** Does the incompatible-changes entry now say that any multi-node `Condition` chain previously wrapped in a single `Expression` node loses its outer parentheses, not only chains that mix AND and OR? - **Validation errors.** Is an `OptionValidationException` thrown while evaluating a `ConditionalRequiredOptions` chain now guarded in `collectErrors`, so the errors already collected are kept? Is the chain still evaluated more than once per rule? - **REST doc.** Has the new sentence moved into the existing Notes list, and is there a multi-node example? Where under `docs/en` is the CLI metadata export it refers to documented? <!-- streview-comment:2445 --> -- 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]
