Github user rdblue commented on the issue: https://github.com/apache/spark/pull/20579 I agree, we should probably add a check for storing a DataFrame with no columns for now. This is normally caught by the pre-insert rules, but since the table is getting "created" in this case there is nothing to check. In the future, I think that the create and insert should be logically separate so that the create will fail and complain that you can't create a table without at least one column. I think it will be cleaner to separate concerns like that.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org