Frank Bax <fbax@sympatico.ca> writes:
> My query produces the correct results for 2005 and 2006; but in 2007 DST
> dates are changing in North America and my query seems to still be using
> 2006 rules for 2007 data.
You certain this is PG 8.1? Every release of 8.1.x has shipped with
timezone data files that included the 2007 rule change. In the 8.0
series, 8.0.4 and up should have that update. PG 7.x releases are
dependent on the operating system's timezone knowledge.
FWIW, my copy of 8.1.5 produces
regression=# select * from dst where tzage <> '12 hrs';
tz1 | tz2 | tzage
------------------------+------------------------+----------
2005-04-03 00:00:00-05 | 2005-04-03 12:00:00-04 | 11:00:00
2005-10-30 00:00:00-04 | 2005-10-30 12:00:00-05 | 13:00:00
2006-04-02 00:00:00-05 | 2006-04-02 12:00:00-04 | 11:00:00
2006-10-29 00:00:00-04 | 2006-10-29 12:00:00-05 | 13:00:00
2007-03-11 00:00:00-05 | 2007-03-11 12:00:00-04 | 11:00:00
2007-11-04 00:00:00-04 | 2007-11-04 12:00:00-05 | 13:00:00
(6 rows)
where
regression=# show timezone;
TimeZone
------------------
America/New_York
(1 row)
(Another possibility I guess is you have the wrong timezone setting?)
regards, tom lane