Re: Using xmin to identify last modified rows

Richard Broersma <richard.broersma@gmail.com> writes:
> On Wed, Feb 25, 2009 at 5:21 AM, Gregory Stark <stark@enterprisedb.com> wrote:
>> I'm not sure using xmin is such a great idea really. It's handy for ad-hoc
>> queries but there are all kinds of cases where it might not give you the
>> results you expect.

> Its been a while since the following emails were written.  Has the
> treatment of xmin changed since then, or is using a timestamp a better
> practice?

> http://archives.postgresql.org/pgsql-novice/2007-02/msg00079.php
> http://archives.postgresql.org/pgsql-hackers/2004-02/msg00654.php

Those statements are all still true, but notice that nowhere do they
suggest doing anything except simple equality comparisons on XIDs.
The OP was looking for ordering, which is a lot trickier, especially
if you might be dealing with old (frozen) tuples.

            regards, tom lane