To: Stephan Szabo <sszabo@megazone.bigpanda.com>
Cc: Ram Nathaniel <ram_nathaniel@hotmail.com>,
	pgsql-sql@postgresql.org
Subject: Re: SQL syntax extentions - to put postgres ahead in the race 
In-reply-to: <20040805220147.O92438@megazone.bigpanda.com> 
References: <BAY7-DAV12V0KUvkRkZ000f7e18@hotmail.com>
	<20040805220147.O92438@megazone.bigpanda.com>
Comments: In-reply-to Stephan Szabo <sszabo@megazone.bigpanda.com>
	message dated "Thu, 05 Aug 2004 22:22:33 -0700"
Date: Fri, 06 Aug 2004 03:07:27 -0400
Message-ID: <8583.1091776047@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

Stephan Szabo <sszabo@megazone.bigpanda.com> writes:
> On Thu, 5 Aug 2004, Ram Nathaniel wrote:
>> 2) aggregated concatenation:

> Theoretically, you should be able to do this right now in PostgreSQL with
> user defined aggregates (although you can't pass a second argument
> currently for the separator).

There's nothing particularly stopping us from supporting
multiple-argument aggregates, except a lack of round tuits.
(I suppose we'd want to rethink the syntax of CREATE AGGREGATE,
but otherwise it ought to be pretty straightforward.)

> I believe that an ordered subquery in FROM
> will currently allow you to get an ordered aggregate, or perhaps you'd
> have to turn off hash aggregation, but I think you should be able to get
> it to keep the ordering.

I think you would want to ORDER BY twice:

    SELECT class, list(student) from
	(select class, student from grades order by class, student) ss
    order by class;

It looks like (at least in CVS tip) planner.c will take into account the
relative costs of doing a GroupAgg vs doing a HashAgg and re-sorting,
but I'm too tired to try it right now...

			regards, tom lane