Github user rxin commented on the issue: https://github.com/apache/spark/pull/21721 Given the uncertainty about how this works across batch, streaming, and CP, and given we are still flushing out the main APIs, I think we should revert this, and revisit when the main APIs are done. In general for API design, it is best to flush out the big skeletons first, and then work on filling the gaps. Think about building a house. You build the frame, and put the stud in, the walls, and then do the final finish. You don't start by putting plumbing fixtures in one room when you are still moving the main plumbing lines.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org