Date: Wed, 15 Sep 2004 16:22:01 +0200
From: Stef <svb@ucs.co.za>
To: Andrew Sullivan <ajs@crankycanuck.ca>
Cc: pgsql-sql@postgresql.org
Subject: Re: Preserving column order when recreating table.
Message-Id: <20040915162201.13be5102@svb.ucs.co.za>
In-Reply-To: <20040915131930.GA28290@phlogiston.dyndns.org>
References: <20040915142445.06b60e3e@svb.ucs.co.za>
	<20040915131930.GA28290@phlogiston.dyndns.org>
User-Agent: sillypheed-claws (anti-aliased)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

Andrew Sullivan mentioned :
=> I'm not sure why you want to do the former, but in any case, it's
Because lazy people write inserts without specifying column names.

=> possible by creating a new table which has things the way you want;
=> select all the old data from the old table into the new table (using
=> the column names to get everything in the order you like, of course),

I like this idea, but each database may have a different table definition
for the same table, and if I want to automate this, I need to figure out the 
column names on the fly.

=> and then rename the old table, rename the new table to the old table
=> name, and drop the old table if you like.

I think I've got the solution now. I'll do it in two steps. 
Fist add/drop all the columns that are not there/not supposed to be there,
and in the second step do what you suggested.

Thanks!!

Kind Regards
Stefan