ryukobayashi commented on PR #6794: URL: https://github.com/apache/hive/pull/6794#issuecomment-6077537991
@deniskuzZ > which ones? The four underlying issues I found were: 1. Generated internal aliases such as `rn` and `cow_matched_row_count` could collide with user-defined columns. 2. The per-file `count(*) OVER (PARTITION BY file_path)` window could buffer and spill all matched rows for a data file. 3. Some CoW MERGE predicates were combined without sufficient grouping, which could allow source-only rows to enter the replacement-marker path and produce NULL-path tombstones. 4. CoW MERGE did not have the standard cardinality-violation handling used by the native MERGE rewrite. The first two issues are directly related to the affected-row-count implementation, so I will address them in this PR. The latter two are broader CoW MERGE correctness and rewrite-architecture issues. I have opened HIVE-30147 to track them separately rather than expanding HIVE-30064 further. -- 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]
