Rachelint commented on PR #15591: URL: https://github.com/apache/datafusion/pull/15591#issuecomment-5477626355
> > So perhaps a simpler alternative is to add EmitTo::Last. > > This is an interesting idea -- is your idea to keep batches managed by blocks internally, but not change the group_index API? And then EmitTo::Last ? Or were you imagining coping the EmitTo::Last rows to the output from the inner Vec allocation I want to try the latter, because: - I found the simple copying is not expansive as we think when I experiment other ideas. - According to the analysis of the coding agent, `Emit::Last` can be easily implemented for most `GroupValues` and `GroupsAccumulator` (compare with the totally blocked approach like this pr) - Not like `Emit::First`, `Emit::Last` will be effective for most `GroupValues` and `GroupsAccumulator`(truncate + shrink) -- 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]
