You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Make AggregateFunction take a single argument #444
Is your feature request related to a problem or challenge? Please describe what you are trying to do.
This makes the representation of aggregate functions more correct, makes code dealing with aggregate functions more simple.
Describe the solution you'd like
Instead of Vec<Expr> use Box<Expr>
Describe alternatives you've considered
Keeping it as is.
fyi, the goal of being Vec was to allow multi-argument aggregates. An example is the pearson correlation between two columns (which can currently be implemented via a UDAF)
@jorgecarleitao Ah, yeah I think that makes sense. I'll just take care of it.
I think there is still some code (e.g. in Ballista protobuf I believe) that assumes only 1 argument, so when we are going to support that, it should be adapted.
Activity
jorgecarleitao commentedon May 30, 2021
fyi, the goal of being
Vecwas to allow multi-argument aggregates. An example is the pearson correlation between two columns (which can currently be implemented via a UDAF)Dandandan commentedon May 30, 2021
@jorgecarleitao Ah, yeah I think that makes sense. I'll just take care of it.
I think there is still some code (e.g. in Ballista protobuf I believe) that assumes only 1 argument, so when we are going to support that, it should be adapted.
docs: Add benchmarking guide (apache#444)