Sender: root@sunpine.krs.ru
Message-ID: <37AF774C.F2F43D1C@krs.ru>
Date: Tue, 10 Aug 1999 08:50:20 +0800
From: Vadim Mikheev <vadim@krs.ru>
Organization: OJSC Rostelecom (Krasnoyarsk)
MIME-Version: 1.0
To: Tom Lane <tgl@sss.pgh.pa.us>
CC: Herouth Maoz <herouth@oumail.openu.ac.il>,
	Matthew Hagerty <matthew@venux.net>, pgsql-interfaces@postgreSQL.org
Subject: Re: [INTERFACES] Proper use of Transactions...
References: <23213.934207520@sss.pgh.pa.us>
Content-Type: text/plain; charset=koi8-r
Content-Transfer-Encoding: 7bit

Tom Lane wrote:
> 
> Even if the backend itself crashes, an uncommitted transaction will
> remain uncommitted, since no entry saying it's committed ever gets
> written to pg_log.  Any tuples the dead backend did manage to write out
> will look like not-yet-committed tuples until the end of time.
> 
> BTW, I am not sure whether VACUUM ever realizes it can reclaim those
> uncommitted tuples.  To do that it'd have to make some kind of

It does.

Vadim