Message-ID: <3FB2A37A.6060201@Yahoo.com>
Date: Wed, 12 Nov 2003 16:17:46 -0500
From: Jan Wieck <JanWieck@Yahoo.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US;
	rv:1.4) Gecko/20030624
MIME-Version: 1.0
To: Holger Jakobs <holger@jakobs.com>
Cc: sszabo@megazone.bigpanda.com, pgsql-sql@postgresql.org
Subject: Re: transaction processing after error in statement
References: <200311111625.hABGP4J11142@bg.bib.de>
In-Reply-To: <200311111625.hABGP4J11142@bg.bib.de>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Holger Jakobs wrote:

> Calling a procedure is a statement as well, and it includes all other
> procedures called from this procedure. So the statement level is always
> the statements that were carried out directly in the transaction. If
> anything within one statement fails, the statement was not carried out
> and must not have any effect. It is not important whether the procedure
> was fired by a trigger or called by another procedure.

So you define the smalles unit being one single statement as issued by 
the client application and receiving one single returncode over the 
frontend/backend protocol.

That's almost what people think of as subtransactions. I think if we 
ever implement them, we will have some session setting that lets the 
backend behave like that and your needs will be satisfied.


Jan

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck@Yahoo.com #