szehon-ho commented on code in PR #6771: URL: https://github.com/apache/iceberg/pull/6771#discussion_r1100488895
########## docs/spark-queries.md: ########## @@ -346,6 +346,9 @@ SELECT * FROM prod.db.table.partitions; Note: For unpartitioned tables, the partitions table will contain only the record_count and file_count columns. +Note2: +The output of the above query might differ between having copy-on-write or merge-on-read strategies. E.g. delete files with MOR strategy aren't applyied when producing the content of the partitions metadata table. As a result if you have renamed a partition (by updating the value of a partition column) then you would see both the 'old' and the 'new' one until you do a rewrite of delete/data files. Review Comment: I actually left a comment on the issue, now I am wondering why we don't use metadata only delete in this case, to go to a codepath where we delete the manifests entry rather than avoid writing delete files. I wonder if its because of doing an update That being said, there may be use-cases where we delete enough of the partition using delete files in separate commit, that results in this, but I am not sure if this specific use case should result in this by design and we should put it specifically. -- 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: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
