aggColumnBatched
The batched sibling of aggColumn (ADR-2026-10-24-20): the same slot arithmetic - one aggInto per group, the very statements the one-slot kernels run - appended into ONE builder, so the output is ONE chunk of groupRows.length slots. The one-slot-per-group shape costs a handful of KB per group in minimum builder blocks (measured on the pivot probes at 10k rows: ~2.7-5 KB per slot chunk; 100k pivot cells allocated 272 MB through it), which many-group consumers such as pivot cells cannot afford. GroupBy.agg keeps the one-slot chunks unchanged (its group counts are small); values are identical by construction.
Parameters
agg | the aggregation kernel name |
outName | the output column name |
col | the aggregated column |
groupRows | the per-group row lists |
Returns
the aggregated output column (one chunk).