From: Tom Lane <tgl@sss.pgh.pa.us>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
cc: David Rowley <dgrowleyml@gmail.com>,
        Pavel Stehule <pavel.stehule@gmail.com>,
        Duncan Sands <duncan.sands@deepbluecap.com>,
        pgsql-bugs@lists.postgresql.org
Subject: Re: range_agg extremely slow compared to naive implementation in
 obscure circumstances
In-reply-to: <20230201115741.qfcjbm4jyji4etvn@alvherre.pgsql>
References: <20230201115741.qfcjbm4jyji4etvn@alvherre.pgsql>
Comments: In-reply-to Alvaro Herrera <alvherre@alvh.no-ip.org>
	message dated "Wed, 01 Feb 2023 12:57:41 +0100"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1367181.1675263071.1@sss.pgh.pa.us>
Date: Wed, 01 Feb 2023 09:51:11 -0500
Message-ID: <1367182.1675263071@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1367182.1675263071%40sss.pgh.pa.us>
Precedence: bulk

Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> On 2023-Jan-31, David Rowley wrote:
>> It might be better if we had multirange_canonicalize() deserialize
>> these once and used some representation that could more easily be
>> qsorted. I'm not planning on doing any work on it though.

> Yeah, maybe it would be possible to have an in-memory representation
> that doesn't require any deparsing, and keep the compact representation
> to be used only for in-data-page storage.  How to do this within the
> constraints of the Datum abstraction is not clear to me.

Perhaps the "expanded datum" mechanism would serve?

src/include/utils/expandeddatum.h

It might be too heavyweight for this application, but I'm not sure.

			regards, tom lane