jacques-n commented on pull request #10934:
URL: https://github.com/apache/arrow/pull/10934#issuecomment-907719679


   >> I think it would be useful to be able to serialize and send e.g. 
sum($expr_0) / mean($expr_1) (with these expressions being possibly unbound to 
a particular table schema) without having to build an aggregation relational 
operator
   
   I'm assuming you mean that you want to avoid having to build a project on 
top of an aggregate? (please confirm my interprestation of what you said).
   
   My initial intuition is that supporting arbitrary expressions in aggregation 
creates more complexity (and heavier requirement on semantic analysis to 
confirm plan validity). I agree that there are situations where you might want 
a compound relational operation that does an aggregation calculation followed 
immediately by a non-aggregate calculation (e.g. the division in your 
expression). However, I think that "compound aggregate" would be an additional 
relational operator we could introduce at a later stage as opposed to being one 
of the initial primitives (or possibly even be an internal concern/optimization 
of a particular execution engine).


-- 
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]


Reply via email to