Tom,
> How so? If you think that the timestamp-without-zone is relative to GMT
> rather than your local zone, you say something like
> extract(epoch from (timestampvar AT TIME ZONE 'GMT'))
Ah, that didn't seem to work before. I must have done the parens wrong.
> Quite honestly, you should be using timestamp WITH time zone for such an
> application anyway. The timestamp without zone datatype is very
> strongly biased towards the assumption that the value is in your local
> timezone, and if you've actually got multiple possible settings of
> TimeZone then it's simply a great way to shoot yourself in the foot.
Well, I was thinking about this on the way to my office this AM, and realized
that there's a fundamental gulf between timestamp-as-real-moment-in-time (the
SQL timestamp and postgres timestamp) and timestamp-as-mark-on-the-calendar
(what I'm dealing with), and that my trouble stems from trying to coerce the
first into the second.
Maybe it's time to hack a datatype ...
--
--Josh
Josh Berkus
Aglio Database Solutions
San Francisco