Message-ID: <429F3B21.9020209@archonet.com>
Date: Thu, 02 Jun 2005 18:00:17 +0100
From: Richard Huxton <dev@archonet.com>
User-Agent: Mozilla Thunderbird 1.0.2 (X11/20050317)
MIME-Version: 1.0
To: Bruno Wolff III <bruno@wolff.to>
Cc: =?ISO-8859-1?Q?K=D6PFERL_Robert?= <robert.koepferl@sonorys.at>,
	"'dpandey@secf.com'" <dpandey@secf.com>,
	pgsql-general@postgresql.org, 'PostgreSQL' <pgsql-sql@postgresql.org>
Subject: Re: [SQL] index row size 2728 exceeds btree maximum, 27
References: <ED4E30DD9C43D5118DFB00508BBBA76EB16736@neptun.sonorys.at>
	<429EFE55.3050601@archonet.com> <20050602163542.GA23150@wolff.to>
In-Reply-To: <20050602163542.GA23150@wolff.to>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Bruno Wolff III wrote:
> On Thu, Jun 02, 2005 at 13:40:53 +0100,
>   Richard Huxton <dev@archonet.com> wrote:
> 
>>Actually, Dinesh didn't mention he was using this for the speed of 
>>lookup. He'd defined the columns as being the PRIMARY KEY, presumably 
>>because he feels they are/should be unique. Given that they are rows 
>>from a logfile, I'm not convinced this is the case.
> 
> 
> Even for case you could still use hashes. The odds of a false collision
> using SHA-1 are so small that some sort of disaster is more likely.
> Another possibility is if there are a fixed number of possible messages,
> is that they could be entered in their own table with a serail PK and
> the other table could reference the PK.

Certainly, but if the text in the logfile row is the same, then hashing 
isn't going to make a blind bit of difference. That's the root of my 
concern, and something only Dinesh knows.

--
   Richard Huxton
   Archonet Ltd