To: Bruce Momjian <bruce@momjian.us>
cc: josh@agliodbs.com, pgsql-core@postgresql.org,
	Heikki Linnakangas <heikki@enterprisedb.com>,
	Alvaro Herrera <alvherre@commandprompt.com>,
	Simon Riggs <simon@2ndquadrant.com>, pgsql-hackers@postgresql.org
Subject: Re: [CORE] FOR SHARE vs FOR UPDATE locks 
In-reply-to: <200612011916.kB1JGRx16905@momjian.us> 
References: <200612011916.kB1JGRx16905@momjian.us>
Comments: In-reply-to Bruce Momjian <bruce@momjian.us>
	message dated "Fri, 01 Dec 2006 14:16:27 -0500"
Date: Fri, 01 Dec 2006 14:27:13 -0500
Message-ID: <1582.1165001233@sss.pgh.pa.us>
From: Tom Lane <tgl@sss.pgh.pa.us>

Bruce Momjian <bruce@momjian.us> writes:
> Josh Berkus wrote:
>> So, I think it needs to go on the list for 8.2.1 or 8.3 (depending on what 
>> changes the fix requires) but I don't think we should hold up the release.

> We cannot add something this major in a minor release --- it would have
> to be 8.3.

If someone thinks of a brilliant solution that doesn't change on-disk
layout, maybe we could implement it in 8.2.x, but right now I'm not
feeling hopeful about that.

The best idea I have at the moment is that we might be able to do
something as part of the proposed plan to fold cmin/cmax into a single
field.  The thought there was that there could be some in-memory state
for tuples that had been modified multiple times by a single xact ---
perhaps that could be extended to cover this problem.  This is just
handwaving at the moment though.  In particular, is-the-tuple-locked-or-not
seems like it has to be externally visible state, so maybe it can't work.

			regards, tom lane