xumingming commented on PR #59402: URL: https://github.com/apache/spark/pull/59402#issuecomment-6187820535
Thanks for the thorough review, @cloud-fan! 1. Keep normalization only when the distinct-group count drops, so a shared IF with SUM/AVG and a repeated conditional count stay off Expand. 2. Gate the union-of-filters guard on the same config, restoring the raw Expand key for user-written FILTER distincts when it is off. 3. Require canEvaluateUnconditionally on every guarded child, so a filtered division stays raw and still raises ANSI divide-by-zero. 4. Update the normalization scaladoc to describe the guarded evaluation domain. 5. Replace the literal-NULL condition test with a nullable column predicate that reaches the guard. 6. Make the plan helper's FILTER assertion require a user-condition reference rather than the gid filter alone. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
