andygrove commented on issue #424: URL: https://github.com/apache/arrow-datafusion/issues/424#issuecomment-848018100
I agree that doing this at the logical plan level is simpler. My concern is that later on, if/when we make Ballista viable, we may want to dynamically optimize the query while it is executing, based on statistics from completed stages, and I was hoping to avoid Spark's approach of going back to the logical plan for this, and just optimize the physical plan. However, we can always revisit this later. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
