On Wednesday, January 25, 2012 7:37:27 am hubert depesz lubaczewski wrote:
> On Wed, Jan 25, 2012 at 10:35:47AM -0500, Tom Lane wrote:
> > hubert depesz lubaczewski <depesz@depesz.com> writes:
> > > Why aren't the 3rd date_parts the same in both cases? I mean - I see
> > > that they are adjusted due to timezone, but why is it happening?
> >
> > Given a timestamp without time zone, timestamp_part('epoch') assumes
> > that it is in session timezone, and rotates it back to UTC so as to
> > satisfy the expectation that epoch values start from zero at midnight
> > UTC. In short, the calculation you're showing does the zone correction
> > an extra time. Don't do that.
>
> ok.
> how can I then have immutable epoch for given point in time?
>
> I thought that this is what I will achieve with extract(epoch from now()
> at time zone 'UTC') but clearly it doesn't work.
> So what options do I have?
Isn't extract(epoch from now()) getting what you want?
>
> Best regards,
>
> depesz
--
Adrian Klaver
adrian.klaver@gmail.com