Message-ID: <4819FFFA.40601@fastmail.net>
Date: Fri, 02 May 2008 01:38:02 +0800
From: Darren Reed <darrenr+postgres@fastmail.net>
Reply-To: darrenr@fastmail.net
User-Agent: Thunderbird 2.0.0.12 (Windows/20080213)
MIME-Version: 1.0
To: Alvaro Herrera <alvherre@CommandPrompt.com>
CC: =?ISO-8859-1?Q?Devrim_G=DCND=DCZ?= <devrim@CommandPrompt.com>,
	pgsql-admin@postgresql.org, Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re: Some additions to pg_filedump...
References: <1208778681.11353.1248996817@webmail.messagingengine.com>
	<1208809188.18775.86.camel@localhost.localdomain>
	<1208944558.8724.1249416347@webmail.messagingengine.com>
	<20080423132049.GF4768@alvh.no-ip.org>
In-Reply-To: <20080423132049.GF4768@alvh.no-ip.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

  Alvaro Herrera wrote:
> Darren Reed wrote:
>
>   
>> I saw this, but when I went in search of activity, there was nothing.
>>
>> Look in:
>> http://sources.redhat.com/ml/rhdb/
>> http://sources.redhat.com/ml/rhdb-announce/
>> http://sources.redhat.com/ml/rhdb-cvs/
>>
>> All of the archives for 2008 are empty...
>>
>> It looks like "PostgreSQL - Red Hat Edition Project" is a defunct
>> project.
>>     
>
> Perhaps it would be apropos for RedHat to donate pg_filedump under BSD
> license so we could include it in contrib?  It's not like they're
> getting much benefit of it being under GPL.
>   

Whether or not that was a hint, I sat down and wrote a program from scratch
to try and dump out only the data files.  I suppose this might be a 
version 1.0.

http://coombs.anu.edu.au/~avalon/pg_dumpdata.tgz

Alas, there is a catch...

I couldn't work out the binary format for all of the data types - as an 
exercise
I tried to reverse engineer their behaviour from examining what got 
stored and
for most of the simple types, I think I've got it right.  I couldn't 
fathom "time";
the geometric ones I couldn't easily find examples of how to insert data 
- I am
assuming the web documentation is self explanatory to those who use it; 
"double"
wasn't a straight fit of "double' from unix (huh?) and "bit" wants a 
length to work
which was against what I was trying to achieve.

It might be interesting to write something that tries to deconstruct the 
index files,
but to me there's no value in that - the value is in the data.

Cheers,
Darren