To: "Decibel!" <decibel@decibel.org>
Cc: "Stefan Arentz" <stefan.arentz@gmail.com>,  <pgsql-sql@postgresql.org>
Subject: Re: Speeding up schema changes
In-Reply-To: <FC6AC0ED-3C3B-41F4-A6DA-75C1F0B34042@decibel.org> (Decibel!'s
	message of "Wed\, 19 Sep 2007 10\:04\:50 -0500")
User-Agent: Gnus/5.11 (Gnus v5.11) Emacs/22.1 (gnu/linux)
References: <adf34c300709030422p7a5bf9ban915c01547c3cb008@mail.gmail.com>
	<874pic0way.fsf@oxford.xeocode.com>
	<FC6AC0ED-3C3B-41F4-A6DA-75C1F0B34042@decibel.org>
From: Gregory Stark <stark@enterprisedb.com>
Organization: EnterpriseDB
Date: Wed, 19 Sep 2007 18:43:57 +0100
Message-ID: <87bqbyim8y.fsf@oxford.xeocode.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

"Decibel!" <decibel@decibel.org> writes:

> On Sep 3, 2007, at 7:26 AM, Gregory Stark wrote:
>> Also, incidentally do you have a good reason to use CHAR instead of  varchar
>> or
>> text? char(64) will take 64 bytes (actually 68 bytes in 8.2) even  if you
>> don't
>> store anything more in it. text or varchar will take only as many  bytes as
>> the
>> data you're storing (plus 4 bytes).
>
> Hrm, do we actually pad before storing? ISTM we should really do that  the
> other way around...

Yes we do. And it isn't really fixable either. The problem is the familiar old
problem that in Postgres the typmod is not really part of the type and not
always available when we need it to interpret the datum.

-- 
  Gregory Stark
  EnterpriseDB          http://www.enterprisedb.com