Message-ID: <452F7CFC.50009@logix-tt.com>
Date: Fri, 13 Oct 2006 13:48:12 +0200
From: Markus Schaber <schabi@logix-tt.com>
Organization: Logical Tracking and Tracing International AG, Switzerland
User-Agent: Thunderbird 1.5.0.7 (X11/20060927)
MIME-Version: 1.0
To: Daniel Drotos <drdani@mazsola.iit.uni-miskolc.hu>
Cc: pgsql-sql@postgresql.org
Subject: Re: deleting rows in specific order
References: <Pine.LNX.4.61.0610092207210.24332@mazsola.iit.uni-miskolc.hu>
	<452CE81A.5010001@logix-tt.com>
	<Pine.LNX.4.61.0610112108100.16984@mazsola.iit.uni-miskolc.hu>
In-Reply-To: <Pine.LNX.4.61.0610112108100.16984@mazsola.iit.uni-miskolc.hu>
Content-Type: text/plain; charset=ISO-8859-15
Content-Transfer-Encoding: 7bit

Hi, Daniel,

Daniel Drotos wrote:

>> What is it that you're trying to achieve with the ordered delete? Maybe
>> there's a better way to solve your underlying problem.
> 
> I've a table which has a field for some info (number) and a date field
> (insertion date) and others.
> 
> After inserting new rows into this table, very old rows are going to be
> deleted. But value of deleted info should be remembered somehow, so
> there is a trigger on delete which keeps sum of the deleted info values
> in a separate table. This makes the application happy.
> 
> Trigger's algorithm is a little bit more dificult which I would be able
> to simplify if rows are deleted in date order.

Ah, I see.

I don't know exactly what you're doing here, but at least for a sum, the
order of summed values is not relevant.

And due to the transaction system, other applications won't see
inconsistent intermediate states, as the whole DELETE run and all
trigger calls are part of the same transaction.

> Now I'm doing it in a plpgsql function using a `for in select loop' to
> delete rows one by one.

I'm afraid that this is the only solution if you can't simplify the
trigger by other means.

HTH,
Markus
-- 
Markus Schaber | Logical Tracking&Tracing International AG
Dipl. Inf.     | Software Development GIS

Fight against software patents in Europe! www.ffii.org
www.nosoftwarepatents.org