Aggregates

DuckDB’s aggregate functions, exposed as free functions (defined with #[aggregate]) in better_duck_diesel::dsl::*. Use them in a grouped or aggregating .select.

use diesel::prelude::*;
use better_duck_diesel::dsl::*;

let m: Option<f64> = products::table.select(median(products::price)).first(&mut conn)?;

Diesel core already ships count/sum/avg/min/max (in diesel::dsl) — this crate adds only what core lacks. Most aggregates return Nullable<…> because an empty group yields NULL; load into an Option.

Watch DuckDB’s result widening on the core aggregates: SUM over an INTEGER column widens to HUGEINT, which the BigInt adapter rejects. Sum a DOUBLE column, or CAST the result. See Select → counting.

General

Aggregate	Signature	Returns
`bool_and(x)`	`Bool`	`Nullable<Bool>`
`bool_or(x)`	`Bool`	`Nullable<Bool>`
`string_agg(x, sep)`	`Text, Text`	`Nullable<Text>`
`arg_max(arg, val)`	`Text, Double`	`Nullable<Text>`
`arg_min(arg, val)`	`Text, Double`	`Nullable<Text>`
`product(x)`	`Double`	`Nullable<Double>`
`any_value(x)`	`Text`	`Nullable<Text>`
`array_agg(x)`	`Text`	`DuckList`

let names: Option<String> = products::table
    .select(string_agg(products::name, ", "))
    .first(&mut conn)?;

array_agg returns a DuckList, which the typed loader cannot read — route it through diesel::sql_query + QueryableByName. See Type mapping.

Statistical

Double -> Nullable<Double> unless noted:

Dispersion & shape

stddev_samp, stddev_pop, var_samp, var_pop, median, skewness, kurtosis, entropy, mad, mode.

Bivariate & regression

corr(y, x), covar_samp(y, x), covar_pop(y, x), regr_slope(y, x), regr_intercept(y, x), regr_r2(y, x), regr_count(y, x) (→Nullable<BigInt>).

Approximate & quantiles

Aggregate	Signature	Returns
`approx_count_distinct(x)`	`Text`	`BigInt`
`approx_quantile(x, pos)`	`Double, Float`	`Nullable<Double>`
`reservoir_quantile(x, q)`	`Double, Double`	`Nullable<Double>`
`quantile_cont(x, pos)`	`Double, Double`	`Nullable<Double>`
`quantile_disc(x, pos)`	`Double, Double`	`Nullable<Double>`

Bitwise

BigInt -> Nullable<BigInt>: bit_and(x), bit_or(x), bit_xor(x).

JSON

json_group_array(x: Text) (→Text) aggregates values into a JSON array; json_group_object(key, value) (both Text, →Text) aggregates key/value pairs into a JSON object. See JSON functions for the scalar side.

Custom aggregate UDFs

Your own aggregates registered with #[duckdb_aggregate] are callable the same way once declared with define_sql_function! + #[aggregate]. See Calling your own UDFs and the UDF macros.

Select

Grouped selects and GROUP BY.

JSON functions

Scalar JSON helpers.

Calling your own UDFs

Type mapping

Reading DuckList aggregate results.

Getting Started

CRUD

DuckDB SQL DSL

Types & Pooling

General

Statistical

Approximate & quantiles

Bitwise

JSON

Custom aggregate UDFs

See also

Select

JSON functions

Calling your own UDFs

Type mapping

​General

​Statistical

​Approximate & quantiles

​Bitwise

​JSON

​Custom aggregate UDFs

​See also

Select

JSON functions

Calling your own UDFs

Type mapping

General

Statistical

Approximate & quantiles

Bitwise

JSON

Custom aggregate UDFs

See also