Re: Add proper planner support for ORDER BY / DISTINCT aggregates

On Wed, Jul 27, 2022 at 6:46 AM David Rowley <dgrowleyml@gmail.com> wrote:

> On Tue, 26 Jul 2022 at 19:39, Richard Guo <guofenglinux@gmail.com> wrote:
> > Also I'm wondering if it's possible to take into consideration the
> > ordering indicated by existing indexes when determining the pathkeys. So
> > that for the query below we can avoid the Incremental Sort node if we
> > consider that there is an index on t(a, c):
> >
> > # explain (costs off) select max(b order by b), max(c order by c) from t
> group by a;
> >                  QUERY PLAN
> > ---------------------------------------------
> >  GroupAggregate
> >    Group Key: a
> >    ->  Incremental Sort
> >          Sort Key: a, b
> >          Presorted Key: a
> >          ->  Index Scan using t_a_c_idx on t
> > (6 rows)
>
> That would be nice but I'm not going to add anything to this patch
> which does anything like that. I think the patch, as it is, is a good
> meaningful step forward to improve the performance of ordered
> aggregates.


Concur with that.


> There are other things in the planner that could gain from what you
> talk about. For example, choosing the evaluation order of WindowFuncs.
> Perhaps it would be better to try to tackle those two problems
> together rather than try to sneak something half-baked along with this
> patch.


That makes sense. The patch looks in a good shape to me in this part.

Thanks
Richard