alamb commented on code in PR #11371:
URL: https://github.com/apache/datafusion/pull/11371#discussion_r1676414149
##########
datafusion/expr/src/planner.rs:
##########
@@ -161,6 +162,28 @@ pub trait ExprPlanner: Send + Sync {
) -> Result<PlannerResult<Vec<Expr>>> {
Ok(PlannerResult::Original(args))
}
+
+ /// Plans a `RawAggregateUDF` based on the given input expressions.
+ ///
+ /// Returns a `PlannerResult` containing either the planned aggregate
function or the original
+ /// input expressions if planning is not possible.
+ fn plan_aggregate_udf(
+ &self,
+ aggregate_function: RawAggregateUDF,
+ ) -> Result<PlannerResult<RawAggregateUDF>> {
+ Ok(PlannerResult::Original(aggregate_function))
Review Comment:
> For sql, count(), count(1) and count(*) are equivalent things that can all
to one single Expr, so I think it is possible to have the conversion in planner.
I agree they are equivalent things. But I am thinking it may be hard to
teach everyone who builds `Exprs` to call `count_star` rather than
`count(lit(1))` or `count(Expr::Wildcard)`
To be clear I don't have a massively strong opinion, but it seems like since
we don't control people's creation of `Expr` we aren't going to be able to
prevent stuff like `count(1)`
--
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]
