Nikk8091 commented on PR #12944: URL: https://github.com/apache/seatunnel/pull/12944#issuecomment-6474561399
Thanks @SEZ9 and @DanielLeens. Pushed `eba1ccb9` with documentation/Javadoc changes only: - Corrected both precedence statements in each EN/ZH Metadata SPI document: fetched datasource values override conflicting job values. Runtime precedence is unchanged. - Documented the key contract on `MetadataProvider.datasourceMap`, `resolveDataSourceConfigs`, `mergeConfig`, and in both SPI documents: unwrapping loses quoting, invalid-path keys are protected by the shared helper, valid `a->b` paths remain nested, and dots remain literal. This does not change provider-key semantics. - Left the optional full-job round-trip refactor for a separate follow-up, as suggested, to avoid changing the preservation fix while its shared helper is still unmerged. I have not changed provider-property parsing or introduced a competing helper. Coverage clarification: the two metadata-disabled parameter rows check the unchanged bypass behavior, not the repaired reparse boundary. The metadata-enabled no-datasource rows and the four datasource-merge rows exercise the rebuild paths. The prior combined-tree results remain historical results against helper head `3d6c855f`, not validation of the latest prerequisite or this standalone head. `mvn -o -pl seatunnel-api spotless:apply` and `git diff --check` passed. I also attempted module `verify -DskipTests`; it stops at the expected unresolved `ConfigMapPathUtils` import. No new passing test or CI result is claimed. #12443 is still open, so this PR remains draft. Once it merges, I will update to that base and rerun the metadata unit tests and existing `MetalakeIT.testDataSource` in CI. -- 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]
