Subject: Re: definative way to place secs from epoc into timestamp
From: Bret Hughes <bhughes@elevating.com>
To: postgresql sql list <pgsql-sql@postgresql.org>
In-Reply-To: <13089.1109793126@sss.pgh.pa.us>
References: <1109780719.28180.86.camel@bretsony>
	<13089.1109793126@sss.pgh.pa.us>
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Date: 02 Mar 2005 14:07:46 -0600
Message-Id: <1109794066.27857.100.camel@bretsony>
Mime-Version: 1.0

On Wed, 2005-03-02 at 13:52, Tom Lane wrote:
> Bret Hughes <bhughes@elevating.com> writes:
> > I give up.  I have STFW and STFM and still do not feel like I have a
> > good way to update/insert into a timestamp w/o TZ column with an integer
> > representing seconds from epoch.
> 
> The docs say:
> 
>   Here is how you can convert an epoch value back to a time stamp: 
> 
>      SELECT TIMESTAMP WITH TIME ZONE 'epoch' + 982384720 * INTERVAL '1 second';
> 
> If you want a timestamp w/o time zone then the right thing depends on
> what you think the reference epoch is.  If you do
> 
>      SELECT TIMESTAMP WITHOUT TIME ZONE 'epoch' + 982384720 * INTERVAL '1 second';
> 
> then what you will get is the correct equivalent of the Unix timestamp
> in GMT time.  If you do the first calculation and then cast to timestamp
> w/o time zone then what you will get is a correct equivalent in your
> TimeZone setting.  For instance


Thanks for the feed back tom  I say that but I could not believe that I
have to jump through all those hoops on an insert or update

update mytable set (lasttime =(SELECT TIMESTAMP WITH TIME ZONE 'epoch' +
982384720 * INTERVAL '1 second') )

is this what you are saying I need to do?

also, what is happening with abstime(982384720)?  this works as expected
(by me ).  Is this a bad idea?  I can't believe that all the complicated
string manipulation stuff is there but I have to run a subselect to
insert a numeric value that I suspect is close to how it is stored
anyway.  Of course the last part is a WAG.


Bret