Message-ID: <43430EEB.80500@archonet.com>
Date: Wed, 05 Oct 2005 00:23:23 +0100
From: Richard Huxton <dev@archonet.com>
User-Agent: Mozilla Thunderbird 1.0.6 (X11/20050912)
MIME-Version: 1.0
To: "Jim C. Nasby" <jnasby@pervasive.com>
Cc: Scott Marlowe <smarlowe@g2switchworks.com>,
	Tom Lane <tgl@sss.pgh.pa.us>, fmiddleton@verizon.net,
	pgsql-sql@postgresql.org
Subject: Re: Why doesn't the SERIAL data type automatically have a
References: <4338961E.20100@verizon.net> <5899.1127782998@sss.pgh.pa.us>
	<1127835194.29347.12.camel@state.g2switchworks.com>
	<20051004225250.GN40138@pervasive.com>
In-Reply-To: <20051004225250.GN40138@pervasive.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Jim C. Nasby wrote:
>>>>Is there some reason why the SERIAL data type doesn't automatically have 
>>>>a UNIQUE CONSTRAINT.
>>>
>>>It used to, and then we decoupled it.
[snip]
> Arguably it would have been better to make the default case add either
> UNIQUE or PRIMARY KEY with a way to over-ride.

Arguably SERIAL shouldn't be a type at all since it's nothing to do with 
defining a set of values. If you were being clean about it you'd have to 
have something like "mycol INTEGER SERIAL UNIQUE", then wire SERIAL to a 
generator function for the type in question.

> If newbies are getting burned maybe it would be useful to toss a NOTICE
> or maybe even WARNING when a serial is created without a unique
> constraint of some kind?

Don't forget the NOT NULL too. Perhaps simpler to have a PGIDENT 
pseudo-type that implies "UNIQUE NOT NULL" and then explain the 
difference in the docs.

--
   Richard Huxton
   Archonet Ltd