From: Josh Berkus <josh@agliodbs.com>
Organization: PostgreSQL @ Sun
To: Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re: VLDB Features
Date: Fri, 14 Dec 2007 20:34:57 -0800
User-Agent: KMail/1.8.2
Cc: Neil Conway <neilc@samurai.com>, Hannu Krosing <hannu@skype.net>,
	pgsql-hackers@postgresql.org, Simon Riggs <simon@2ndquadrant.com>
References: <1197371508.4255.1327.camel@ebony.site>
	<1197677930.1536.18.camel@dell.linuxdev.us.dell.com>
	<16876.1197680636@sss.pgh.pa.us>
In-Reply-To: <16876.1197680636@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="utf-8"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200712142034.57808.josh@agliodbs.com>

Tom,

> I think such an approach is doomed to hopeless unreliability.  There is
> no concept of an error that doesn't require a transaction abort in the
> system now, and that doesn't seem to me like something that can be
> successfully bolted on after the fact.  Also, there's a lot of
> bookkeeping (eg buffer pins) that has to be cleaned up regardless of the
> exact nature of the error, and all those mechanisms are hung off
> transactions.

There's no way we can do a transactionless load, then?  I'm thinking of the 
load-into-new-partition which is a single pass/fail operation.  Would 
ignoring individual row errors in for this case still cause these kinds of 
problems?

-- 
Josh Berkus
PostgreSQL @ Sun
San Francisco