Lukasz Stec
Lukasz Stec
> also allows for partial aggregation to be evaluated. I believe partial aggregation supports `MarkDistinct` too
closing as there is no interest in merging this
Yes, it's fixed by https://github.com/trinodb/trino/pull/15927
the ``` Aggregation[FINAL](group)[aggr := count(aggr_partial), aggr_distinct := count(aggr_distinct_partial)] Exchange[group] Aggregation[PARTIAL](group)[aggr_partial := count(a), aggr_distinct_partial := count(distinct b)] Exchange[b] ``` rewrite looks like a potential good optimization and it's still not done...
> I think that this is an overflow when merging OperatorStats That is unlikely given the positions are kept in `long`.
> @lukasz-stec is inputPositions == 0 a valid value? Yes it is