Date: Fri, 7 Apr 2023 22:34:08 -0400
From: Bruce Momjian <bruce@momjian.us>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: "Fujii.Yuki@df.MitsubishiElectric.co.jp" <Fujii.Yuki@df.mitsubishielectric.co.jp>,
	Andres Freund <andres@anarazel.de>,
	Alexander Pyhalov <a.pyhalov@postgrespro.ru>,
	Tomas Vondra <tomas.vondra@enterprisedb.com>,
	PostgreSQL-development <pgsql-hackers@postgresql.org>,
	Julien Rouhaud <rjuju123@gmail.com>,
	Daniel Gustafsson <daniel@yesql.se>,
	Ilya Gladyshev <i.gladyshev@postgrespro.ru>
Subject: Re: Partial aggregates pushdown
Message-ID: <ZDDSoFv4KZkGmwQk@momjian.us>
References: <c4e769a611278cb989e7832cb8577ab1@postgrespro.ru>
 <OS3PR01MB6660CB590AEF3644A9EEE07D95189@OS3PR01MB6660.jpnprd01.prod.outlook.com>
 <20221207185943.g3pkqseovvrk2n2e@awork3.anarazel.de>
 <OS3PR01MB6660A1DAF6619B9A0BF7D39B95E19@OS3PR01MB6660.jpnprd01.prod.outlook.com>
 <ZCYeIeFRgL2uBeuk@momjian.us>
 <OS3PR01MB6660B900F33523AAD0370D62958F9@OS3PR01MB6660.jpnprd01.prod.outlook.com>
 <ZC95C0+PVhVP3iax@momjian.us>
 <OSZPR01MB6662875933E87F37C711B4FC95969@OSZPR01MB6662.jpnprd01.prod.outlook.com>
 <ZDDIU2fJxZm/k5kj@momjian.us>
 <3278991.1680918900@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <3278991.1680918900@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/ZDDSoFv4KZkGmwQk%40momjian.us>
Precedence: bulk

On Fri, Apr  7, 2023 at 09:55:00PM -0400, Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > What I don't want is an error-prone setup where administrators have to
> > remember what the per-server settings are.  Based on your suggestion,
> > let's allow CREATE SERVER to have an option 'enable_async_aggregate' (is
> > that the right name?), which defaults to 'true' for _all_ servers, even
> > those that don't support async aggregates.
> 
> Uh, what?  Why would we not be able to tell from the remote server's
> version number whether it has this ability?

That was covered here:

	https://www.postgresql.org/message-id/ZC95C0%2BPVhVP3iax%40momjian.us

	I think we have three possible cases for aggregate pushdown to FDWs:

	1)  Postgres built-in aggregate functions
	2)  Postgres user-defined & extension aggregate functions
	3)  aggregate functions calls to non-PG FDWs

	Your patch handles #1 by checking that the FDW Postgres version is the
-->	same as the calling Postgres version.  However, it doesn't check for
-->	extension versions, and frankly, I don't see how we could implement that
-->	cleanly without significant complexity.

The issue is not a mismatch of postgres_fdw versions but the extension
versions and whether the partial aggregate functions exist on the remote
side, e.g., something like a PostGIS upgrade.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Embrace your flaws.  They make you human, rather than perfect,
  which you will never be.