On Thu, 2011-06-02 at 20:28 -0400, Robert Haas wrote:
> But that doesn't seem like enough, because if someone adds '1 day',
> knowing the offset isn't sufficient to figure out the answer. You
> have to know where the DST boundary is.
Good point, I guess the timezone itself needs to be stored. That's a
little unfortunate, because timezones are somewhat of a moving target
(which I think was Tom's point).
That means that we'd need an entire history (and future?) of timezone
definitions, and apply the timezone definition as of the associated
timestamp to get the offset. Or, should we apply the timezone definition
as of the "real" time the value was entered?
Regards,Jeff Davis