From: Peter Eisentraut <peter_e@gmx.net>
To: Andreas <maps.on@gmx.net>, pgsql-sql@postgresql.org
Subject: Re: Notation of index
Date: Wed, 7 Jan 2004 19:36:53 +0100
User-Agent: KMail/1.5.1
References: <3FFC3D30.2080508@gmx.net>
In-Reply-To: <3FFC3D30.2080508@gmx.net>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200401071936.53243.peter_e@gmx.net>

Andreas wrote:
> Why can['t] we use INDEX the same way as UNIQUE ?
> Perhaps even as in
> ...
>   numba   INT4    NOT NULL    INDEX
> ...

If the choice were just index or no index, then this might be ok, but 
when you create an index you can choose the index type, the operator 
class, a partial index predicate, so the notation would quickly become 
complicated.  You can also consider similar arguments about why rules 
and triggers are created separately.

What would clearly be useful is a way to automatically generate names 
for indexes.  Maybe we should think about implementing something for 
that.