milenkovicm commented on PR #1722: URL: https://github.com/apache/datafusion-ballista/pull/1722#issuecomment-4482812798
> Wasted memory: a downstream aggregate stays Linear (hash table, O(distinct-groups) memory) even when an upstream stage completes and grants ordering on the group-by columns — where Sorted (streaming, O(1) memory) would suffice. In which cases will this happen? When can w guarantee preserving sorting across shuffles ? > Stale correctness assumption: after a subtree rewrite, an aggregate may hold a Sorted claim when its input is no longer ordered. Can you please give an example when would this claim be true? -- 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]
