Support complete distinct usage for aggregate expressions

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
Here are multiple aggregate expressions have not support distinct usage yet.
Though optimizer SingleDistinctToGroupBy (#1315) supports single distinct usage, there are more SQL scenes have not been covered, like

SELECT AVG(c1), SUM(DISTINCT c2) FROM test

I would like to take effort to work on the following tickets to complete distinct usage for aggregate expressions

Describe the solution you'd like
Adds specific aggregate expression support, like expressions::DistinctSum for sum(distinct)

Describe alternatives you've considered
No.

Additional context
No.

@andygrove

@andygrove @alamb can you please assign this issue to me? Thank you ❤️

@andygrove

@andygrove @alamb can you please assign this issue to me? Thank you ❤️

Done -- thanks @WinkerDu

Refactor distinct aggregate implementations to use common buffer (#18348

All done

I think there's still some that weren't listed here that still don't support distinct.

Explicitly denies

stddev

datafusion/datafusion/functions-aggregate/src/stddev.rs

Lines 118 to 123 in bb121a8

    
           fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> { 
        
               if acc_args.is_distinct { 
        
                   return not_impl_err!("STDDEV_POP(DISTINCT) aggregations are not available"); 
        
               } 
        
               Ok(Box::new(StddevAccumulator::try_new(StatsType::Sample)?)) 
        
           }

approx_median

datafusion/datafusion/functions-aggregate/src/approx_median.rs

Lines 130 to 135 in bb121a8

    
           fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> { 
        
               if acc_args.is_distinct { 
        
                   return not_impl_err!( 
        
                       "APPROX_MEDIAN(DISTINCT) aggregations are not available" 
        
                   ); 
        
               }

May be missing support

These functions don't explicitly use the distinct flag to handle distinct specific behaviour, but depending on their semantics maybe distinct doesn't make sense for them? Anyway just listing them:

Navigation Menu

Repository navigation

Support complete `distinct` usage for aggregate expressions #2406

Explicitly denies

May be missing support

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

	fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> {
	if acc_args.is_distinct {
	return not_impl_err!("STDDEV_POP(DISTINCT) aggregations are not available");
	}
	Ok(Box::new(StddevAccumulator::try_new(StatsType::Sample)?))
	}

	fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> {
	if acc_args.is_distinct {
	return not_impl_err!(
	"APPROX_MEDIAN(DISTINCT) aggregations are not available"
	);
	}

Support complete distinct usage for aggregate expressions #2406

Description

Activity

WinkerDu commented on May 2, 2022

alamb commented on May 2, 2022

Dandandan commented on Jun 1, 2026

Jefffrey commented on Jun 1, 2026

Explicitly denies

May be missing support

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Support complete `distinct` usage for aggregate expressions #2406