Custom datasets and Table based datasets (ObjectMappedTable, Table, IndexedTable, KeyValueTable) require a custom Hive storage handler to be explored. This lets users explore those datasets through the CDAP explore service. However, it means they cannot be explored directly through the Hive CLI or through other tools like Hue that don't go through the CDAP explore service. Ideally, users should be able to query these datasets directly through Hive, without using the CDAP explore service.
Note: FileSet based datasets can already be explored directly through Hive, as they don't use a custom storage handler.
A few things need to happen to make this work.
1. Create a jar for the custom CDAP storage handler. Users can place this on the Hive classpath to make it available to Hive clients.
2. Investigate what it would take to start the transaction within the input format.