Date: Mon, 7 Jan 2008 11:18:12 -0500
From: Andrew Sullivan <ajs@crankycanuck.ca>
To: pgsql-sql@postgresql.org
Subject: Re: Cheers for DISTINCT ON
Message-ID: <20080107161812.GS18581@crankycanuck.ca>
References: <b6e8f2e80801070802n7b04b0e3lbbec1d284bef1f9f@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <b6e8f2e80801070802n7b04b0e3lbbec1d284bef1f9f@mail.gmail.com>
User-Agent: Mutt/1.5.13 (2006-08-11)

On Mon, Jan 07, 2008 at 05:02:27PM +0100, Peter Kovacs wrote:
> I just wanted to give my cheers for DISTINCT ON. It is a great
> feature, I've just found a really good use for it. I am just wondering
> why it didn't make it into the standards.

Likely because neither Oracle Corp nor IBM nor (at the time, I guess) Sybase
had an implementation they were willing to spend enough time promoting.

> wondering ever since why the worse so often gets the upper-hand over
> the better. (I am obviously having a hard time "growing-up" :-) )

There is a long history in the computing literature on this, but the best
explanation is probably still Richard Gabriel's 'The Rise of "Worse is
Better"': <http://www.jwz.org/doc/worse-is-better.html>.  Note that Gabriel
himself is of two minds about that paper:
<http://www.dreamsongs.com/WorseIsBetter.html>.  I still think it's worth
reading.  Just about everything I've ever written or said turned out to be
wrong in some sense, but it was still usually the result of the best
argument I could make at the time. 

A