On Thursday, August 14, 2014, Jeff Davis <pg...@j-davis.com> wrote:
> On Thu, 2014-08-14 at 10:06 -0400, Tom Lane wrote:
> > If you're following the HashJoin model, then what you do is the same
> > it does: you write the input tuple back out to the pending batch file for
> > the hash partition that now contains key 1001, whence it will be
> > when you get to that partition. I don't see that there's any special
> > here.
> HashJoin only deals with tuples. With HashAgg, you have to deal with a
> mix of tuples and partially-computed aggregate state values. Not
> impossible, but it is a little more awkward than HashJoin.
Not to mention future cases if we start maintaining multiple state
values,in regarded to grouping sets.