alisha-1000 commented on PR #36722: URL: https://github.com/apache/beam/pull/36722#issuecomment-3833800400
Thanks for the detailed analysis — this is very helpful. The hypothesis around AutoValueSchema field ordering during state restoration makes sense, especially given that it only reproduces on job updates (not drain). That aligns with schema-based state decoding paths being exercised only during update. It might be worth confirming whether the temporary `KafkaSourceDescriptor.create()` workaround is still needed, or if removing it (as suggested in #36357) resolves the ordering mismatch in newer Beam versions. Happy to help test or validate a fix once there’s guidance from maintainers. -- 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]
