milenkovicm commented on PR #1647: URL: https://github.com/apache/datafusion-ballista/pull/1647#issuecomment-4365906094
> I’m not sure the current AQE implementation is suitable. It might eagerly repartition both sides of the join. > > We’ll likely need a rule for optimising broadcast join replanning as stages progress and new statistics are gathered. > > To support this, I believe we should step back from the current AQE implementation. Instead of eagerly splitting the entire plan into stages, we should focus on the independent, executable parts of the plan. I’ve considered this as an alternative to the current approach but now I feel we have a compelling reason to do it. > > This should simplfy design making planning and reoptimization simpler updated AQE in #1649 -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
