RE: Partial aggregates pushdown

Hi Mr.Momjian

> First, am I correct?
Yes, you are correct. This patch uses new special aggregate functions for partial aggregate
(then we call this partialaggfunc).

> Second, how far away is this from being committable
> and/or what work needs to be done to get it committable, either for PG 16 or 17?
I believe there are three: 1. and 2. are not clear if they are necessary or not; 3. are clearly necessary.
I would like to hear the opinions of the development community on whether or not 1. and 2. need to be addressed.

1. Making partialaggfunc user-defined function
In v17, I make partialaggfuncs as built-in functions.
Because of this approach, v17 changes specification of BKI file and pg_aggregate.
For now, partialaggfuncs are needed by only postgres_fdw which is just an extension of PostgreSQL.
In the future, when revising the specifications for BKI files and pg_aggregate when modifying existing PostgreSQL functions,
It is necessary to align them with this patch's changes.
I am concerned that this may be undesirable.
So I am thinking that v17 should be modified to making partialaggfunc as user defined function.

2. Automation of creating definition of partialaggfuncs
In development of v17, I manually create definition of partialaggfuncs for avg, min, max, sum, count.
I am concerned that this may be undesirable.
So I am thinking that v17 should be modified to automate creating definition of partialaggfuncs
for all built-in aggregate functions.

3. Documentation
I need add explanation of partialaggfunc to documents on postgres_fdw and other places.

Sincerely yours,
Yuuki Fujii

--
Yuuki Fujii
Information Technology R&D Center Mitsubishi Electric Corporation