ozankabak commented on issue #5076: URL: https://github.com/apache/arrow-datafusion/issues/5076#issuecomment-1520715267
I find Idea #1 reasonable. If we think of `ExecutionPlan`'s as nodes in the computation graph, writer nodes are simply terminal/sink nodes -- but they are still nodes. Therefore some methods like `output_ordering` not being consumed by anyone doesn't seem unnatural to me, this would be analogous to sink nodes not having any outgoing edges. All in all, I think following Spark's approach is reasonable in this case. I am curious to hear what @andygrove thinks about @alamb's points, though. -- 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]
