alamb commented on issue #22395: URL: https://github.com/apache/datafusion/issues/22395#issuecomment-5061039338
> > non-overlapping (is this a enum Distribution...?) would have some really interesting properties. > > [@stuhood](https://github.com/stuhood) some properties like supporting halo rows for partitioned window functions: [#23026](https://github.com/apache/datafusion/pull/23026) I think these would greatly speed up some of our workloads Yeah, it is interesting question of where HALO rows would logically fit -- I do think the natural place to put it is in the range repartition operator -- perhaps an API would look like (emit the first N rows from the last partition)? Also this probably makes sense only when the data is first ordered by the same column it is being range partitioned on (whihc is true for the window plans, but not in general 🤔 ) -- 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]
