Github user mgaido91 commented on the issue: https://github.com/apache/spark/pull/20999 @cloud-fan I checked the feasibility of doing what you suggested, but unfortunately that method is very hard to be used (and I am not sure how safe it is). The main issues are: - it doesn't accept a string for the filters, but its encoding in a byte array. The point is that the encoding is performed by Hive utils and relying on them might be very hard (especially it is hard to track any change Hive makes there across the versions); - a minor issue is that also an integer is required there and I have not been able to understand what it represents and therefore how to use it. I am not sure if then we can go ahead with this solution, even though it is suboptimal and eventually create a new ticket for leveraging the Hive metastore API you mentioned in a separate PR, as it is not going to be trivial to do that. What do you think?
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org