Module aggregate 

SQL Aggregation Executor

Hash-aggregation operator for GROUP BY and aggregate functions.

Supported aggregates: COUNT(*), COUNT(col), COUNT(DISTINCT col), SUM, AVG, MIN, MAX, MEDIAN, STDDEV (sample, n-1, matching R’s sd() and DuckDB’s stddev).

Pipeline

input rows (post-WHERE)
  └─> group keys evaluated per row ──> hash table of group states
        └─> accumulators updated per row
              └─> finalize: one synthesized row per group
                    └─> HAVING filter ─> ORDER BY ─> OFFSET/LIMIT ─> projection

Semantics notes:

• NULL inputs are skipped by all aggregates except COUNT(*) (SQL standard).
• An ungrouped aggregate over zero rows yields exactly one row (COUNT = 0, other aggregates NULL); a grouped aggregate over zero rows yields zero rows.
• Non-aggregate SELECT columns that are not in GROUP BY resolve to the first value seen in the group (lenient mode, like SQLite / MySQL with ONLY_FULL_GROUP_BY disabled).

Enums

AggFn

Recognized aggregate functions.

Functions

compare_values

Total ordering across SochValue for grouping/sorting.

execute_aggregate

Execute aggregation over materialized input rows (already WHERE-filtered).

is_aggregate_query

Returns true if the SELECT needs the aggregation operator.

render_expr_name

Human-readable name for an expression, used for output column naming and canonical aggregate keys.