Github user lokkju commented on the issue: https://github.com/apache/spark/pull/16479 So it essentially compiles each implementation against different spark versions, then *both* bytecodes are included in the final jar? Then reflection to instantiate it. That works, without too much pain. Might go that route, thanks.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org