Re: effective SELECT from child tables

From: Andrew Dunstan <andrew(at)dunslane(dot)net>
To: "Jim C(dot) Nasby" <jnasby(at)pervasive(dot)com>
Cc: Martijn van Oosterhout <kleptog(at)svana(dot)org>, Simon Riggs <simon(at)2ndquadrant(dot)com>, Ilia Kantor <ilia(at)obnovlenie(dot)ru>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: effective SELECT from child tables
Date: 2005-10-01 16:11:20
Message-ID: 433EB528.1000005@dunslane.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Jim C. Nasby wrote:

>On Sat, Oct 01, 2005 at 02:13:09PM +0200, Martijn van Oosterhout wrote:
>
>
>>On Fri, Sep 30, 2005 at 06:30:10PM -0500, Jim C. Nasby wrote:
>>
>>
>>>On Wed, Sep 28, 2005 at 07:25:46PM +0100, Simon Riggs wrote:
>>>
>>>
>>>>Include the Discriminator as a column in A and it will be inherited by
>>>>all A1, A2, A3.
>>>>e.g. concrete_class char(1) not null
>>>>
>>>>
>>><snip>
>>>
>>>
>>>>This will add 1 byte per row in your superclass... and requires no
>>>>
>>>>
>>>I thought char was actually stored variable-length...? I know there's a
>>>type that actually acts like char does on most databases, but I can't
>>>remember what it is off-hand (it should be mentioned in docs 8.3...)
>>>
>>>
>>IIRC, this is the difference between "char" and char(1). The latter is
>>variable length and can store any character per current encoding, hence
>>the variable length. "char" on the other hand is a one byte (presumably
>>ASCII) character. It's used mainly in the system catalogs...
>>
>>
>
>According to the docs, char == char(1).
>
>

The docs also say:

The type "char" (note the quotes) is different from char(1) in that it
only uses one byte of storage. It is internally used in the system
catalogs as a poor-man's enumeration type.

cheers

andrew

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2005-10-01 16:12:09 Re: [PATCHES] Proposed patch for sequence-renaming problems
Previous Message Tom Lane 2005-10-01 16:01:29 Re: [PATCHES] Proposed patch for sequence-renaming problems