skywalker0618 commented on PR #20337: URL: https://github.com/apache/hudi/pull/20337#issuecomment-6222377899
> Thks for the fix. Since this reader was ported from Flink, have you checked whether the latest upstream Flink implementation already handles this deep nested-array case? If so, could we backport the relevant changes while preserving Hudi’s schema-evolution support? I checked the latest Flink master: there's no newer fix to backport. The last changes there are FLINK-35702 and FLINK-36709, and Hudi already has both. Flink handles an array of rows in its ROW allocator by always stepping down two levels and matching fields by position, so it only supports the standard three-level list and would break Hudi's name-based schema evolution. So this PR unwraps in the ARRAY allocator instead, using the same rules as getArrayElementColumn. -- 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]
