cshuo commented on PR #20328: URL: https://github.com/apache/hudi/pull/20328#issuecomment-6432113456
Thanks @nsivabalan! 1. I checked the inline path: [`collectColumnStats`](https://github.com/apache/hudi/blob/b7ef6fd5feb3fc9d85fd6d4e34031d1e88084481/hudi-client/hudi-client-common/src/main/java/org/apache/hudi/io/HoodieInlineLogAppendHandle.java#L178-L195) delegates to `collectColumnRangeMetadata`, which produces stats for each target field and [returns `createEmpty(...)` when no stats were collected](https://github.com/apache/hudi/blob/b7ef6fd5feb3fc9d85fd6d4e34031d1e88084481/hudi-common/src/main/java/org/apache/hudi/metadata/HoodieTableMetadataUtil.java#L357-L365). So delete-only appends already produce empty stats. I’ve added a [MOR regression test](https://github.com/apache/hudi/blob/b7ef6fd5feb3fc9d85fd6d4e34031d1e88084481/hudi-spark-datasource/hudi-spark/src/test/scala/org/apache/spark/sql/hudi/feature/TestDataSkippingQuery.scala#L307-L389) that verifies the inline delete-block path, empty stats for each indexed column, preserved partition stats, and correct query results with data skipping enabled. 2. Opened [apache/hudi#20441](https://github.com/apache/hudi/issues/20441) to track historical incomplete partition statistics. I also updated the PR description with rebuild guidance, explicitly noting that upgrading alone does not repair existing aggregates. -- 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]
