XiaoHongbo-Hope commented on PR #629: URL: https://github.com/apache/paimon-rust/pull/629#issuecomment-5118326066
> * `table_scan.rs:270` first performs the manifest ADD/DELETE merge, followed by row-range pruning at `table_scan.rs:271`, but `table_scan.rs:286` is where `manifest_entries_after_merge` is recorded. Therefore, this field actually records the count “after merge and row-range pruning.” > * Additionally, `table_scan.rs:275` only overwrites `manifest_entries_after_manifest_filters` with the count after merge and pruning when a row range exists; the semantics of this field are inconsistent depending on whether a row range is present or not. > * For example, if 2 live entries remain after a merge and 1 is subsequently removed by a row range, the current `after_merge` value is incorrectly recorded as 1. It is recommended to save the count immediately after the merge, have the number of row ranges expressed separately by `pruned_by_row_ranges`, and add a test that distinguishes between the two phases. Updated -- 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]
