Github user sv71294 commented on the issue: https://github.com/apache/carbondata/pull/2334 @chenliang613 issue is not about the result, in current master filter is getting handled by presto as it is not generated on connector layer, now with this PR it will be generated at connector layer and passed to core for filter blocklet scanning which will reduce the no of rows passed to presto with filter expression else presto will need to filter the row from universal set. You can also refer to JIRA ISSUE CARBONDATA-2516 for clarity.
---