SEZ9 commented on issue #12928: URL: https://github.com/apache/seatunnel/issues/12928#issuecomment-6332517166
Thanks @Nikk8091 — the scope is yours. @hutiefang76 has confirmed there is no competing implementation and that #12443 stays focused on non-metadata loading. Your plan lines up with the earlier finding: `MetadataProviderManager.resolveDataSourceConfigs()` re-parses the unwrapped job config via `ConfigFactory.parseMap()`, so quoted literal schema keys fail with `ConfigException.BadPath` whenever metadata resolution is enabled, even without a `metadata_datasource_id`. Please make sure the fix covers both `parseMap` boundaries you identified (`resolveDataSourceConfigs()` and `mergeConfig()`) so the quoted key survives the full path. For the PR: 1. Extend `MetadataProviderManagerTest` with the quoted regex-key regression plus the metadata-disabled and enabled-without-datasource controls, so the failure is shown to be specific to the metadata resolution path. 2. Cover actual datasource merges in both flat and nested connector configurations, asserting that valid key-path semantics and existing datasource override precedence are unchanged. 3. Reuse the shared helper from #12443 rather than adding a second conversion policy, and mark #12443 as a dependency as @hutiefang76 asked. The regressions can be prepared on dev in the meantime. 4. Keep the change confined to metadata resolution and its tests, as you proposed. Feel free to link the PR here once it is up. <!-- streview-comment:1954 --> -- 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]
