yujun777 commented on PR #65468: URL: https://github.com/apache/doris/pull/65468#issuecomment-4957500200
I removed the earlier regression case from this PR. I evaluated a stream + CTE + change-type regression path, but it cannot stably prove this specific fix through the current `.out` observability. The bug here is that `RecordPlanForMvPreRewrite` records `tmpPlanForMvRewrite` before stream normalization unless `CTE_CHILDREN_REWRITE_JOBS_MV_REWRITE_USED` includes `NormalizeOlapTableStreamScan`. That recorded temp plan only lives in FE memory, and current SQL / `.out` results do not expose it directly. I also tried using MV chosen/not-chosen as a proxy, but that is not reliable for this query shape. After stream normalization, the incremental stream CTE still contains binlog/change-type lowering and may not semantically match, or stably choose, a base-table MV. So such a regression would only be smoke coverage and could still pass before the fix. Because of that, I kept the direct FE UT instead. The new test forces `tmpPlanForMvRewrite` recording and asserts that no `LogicalOlapTableStreamScan` remains in the recorded temp plans. -- 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]
