DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=mime-version:in-reply-to:references:from:date:message-id:subject:to
	:cc:content-type:content-transfer-encoding;
	b=EP4IluDaO4tR4iuMRjlwRHjHq/8rRoVKJ4kxw36JLECkFy5MSwBxlz28ebCg0O0oCT
	fzp12On8vdZN8O/LeMg1/N2Zowp2OGrOHd74Q1LjWFL1Rqn4EzSih0DRO8qSX53kHThC
	GxONAAh/gB+P13vCCxycz0uoi+LV/o9rOgWFQ=
MIME-Version: 1.0
In-Reply-To: <4DB7D7E3.6010403@douglastechnology.co.uk>
References: <4DB7D7E3.6010403@douglastechnology.co.uk>
From: Josh Kupershmidt <schmiddy@gmail.com>
Date: Thu, 28 Apr 2011 21:20:39 -0400
Message-ID: <BANLkTimyUsgA-DvFi+AweMB4XXogUaMhDA@mail.gmail.com>
Subject: Re: boolean states
To: Jack Douglas <jack@douglastechnology.co.uk>
Cc: pgsql-docs@postgresql.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Wed, Apr 27, 2011 at 4:46 AM, Jack Douglas
<jack@douglastechnology.co.uk> wrote:
> =A0 =A0 =A0 =A0The boolean type can have one of only two states: "true" o=
r "false".
> A third state, "unknown", is represented by the SQL null value.
>
> This sounds like an oxymoron to me.

I'm not crazy about that paragraph's confusion between two and three
states either, but..

> Perhaps that sentence should be changed
> to:
>
> =A0 =A0 =A0 =A0The boolean type can have one of three states: "true" or "=
false" and
> "unknown". The third state, "unknown", is represented by the SQL null val=
ue.
> or:
>
> =A0 =A0 =A0 =A0The boolean type can have one of three states: "true" or "=
false" and
> null. The third state, null, represents the logical value "unknown".

I don't think either of these suggested replacements are any better.
First, a boolean column can be declared NOT NULL. Second, I don't like
the idea of misleading people into thinking that NULL is on equal
footing with the other values of a given datatype, particularly as
your first alternative implies.

I'd vote for just ripping out the:
|  A third state, "unknown", is represented by the SQL null value.

sentence entirely. I see no reason why NULL should be talked about in
particular on the page about boolean data types; there are many data
types, any of which might be NULL.

I almost think it would be worthwhile to have a section in the docs on
the (counterintuitive) behaviors of NULL, such as this great post:
<http://thoughts.j-davis.com/2009/08/02/what-is-the-deal-with-nulls/>
-- or maybe just a link to that page.

Josh