Greg Stark <stark@mit.edu> writes:
> The basic idea is to separate the all the properties of the aggregate
> functions except the final function from the final function into a
> separate object. Giving the optimizer the knowledge that multiple
> aggregate functions use the share the same basic machinery and
> semantics for the state is the magic sauce that's a prerequisite for
> the several ideas we were each thinking of.
Why exactly do you need to invent an "aggregate class" concept for this?
Can't the planner just look in pg_aggregate to see that the
sfunc/stype/initcond are the same for two different aggregates?
regards, tom lane