Re: Partial aggregates pushdown

On Wed, Dec 6, 2023 at 3:41 AM Fujii(dot)Yuki(at)df(dot)MitsubishiElectric(dot)co(dot)jp
<Fujii(dot)Yuki(at)df(dot)mitsubishielectric(dot)co(dot)jp> wrote:
> Are you concerned about the hassle and potential human errors of manually adding new partial
> aggregation functions, rather than the catalog becoming bloated?

I'm concerned about both.

> The process of creating partial aggregation functions from aggregation functions can be automated,
> so I believe this issue can be resolved. However, automating it may increase the size of the patch
> even more, so overall, approach#2 might be better.
> To implement approach #2, it would be necessary to investigate how much additional code is required.

Yes. Unfortunately I fear that there is quite a lot of work left to do
here in order to produce a committable feature. To me it seems
necessary to conduct an investigation of approach #2. If the result of
that investigation is that nothing major stands in the way of approach
#2, then I think we should adopt it, which is more work. In addition,
the problems around transmitting serialized bytea blobs between
machines that can't be assumed to fully trust each other will need to
be addressed in some way, which seems like it will require a good deal
of design work, forming some kind of consensus, and then
implementation work to follow. In addition to that there may be some
small problems that need to be solved at a detail level, such as the
HAVING issue. I think the last category won't be too hard to sort out,
but that still leaves two really major areas to address.

--
Robert Haas
EDB: http://www.enterprisedb.com