Date: Sat, 26 Feb 2005 15:16:10 -0600
From: Bruno Wolff III <bruno@wolff.to>
To: Joel Fradkin <jfradkin@wazagua.com>
Cc: pgsql-sql@postgresql.org
Subject: Re: diference in dates in minutes
Message-ID: <20050226211610.GA4472@wolff.to>
Mail-Followup-To: Joel Fradkin <jfradkin@wazagua.com>,
 pgsql-sql@postgresql.org
References: <20050226180108.GB26473@wolff.to>
	<000401c51c3f$b82188d0$797ba8c0@jfradkin>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <000401c51c3f$b82188d0$797ba8c0@jfradkin>
User-Agent: Mutt/1.5.6i

On Sat, Feb 26, 2005 at 15:14:02 -0500,
  Joel Fradkin <jfradkin@wazagua.com> wrote:
> You probably want to convert the dates to timestamps, subtract them to
> get an interval, extract the epoch to get timme in seconds and then divide
> by 60 to get time in minutes.
> 
> The converting date to timestamp part isn't trivial. You need to decide
> on what you mean when you do this. If you really have timestamps in the
> first place, then you can skip the covernsion step.
> 
> They are dates and I did find I could do date - date to give me an interval 
> date_part('epoch',date-date) returns in secs so /60

date - date won't give you an interval, it will give you an integer of some
sort.

> This appeared to work ok without converting to time stamps, but maybe I am
> missing it if it is not correct as the example I looked at was a large
> difference. The app is analyzing Tlogs and the difference should never be
> too large, so I will further analyze it with real data.
> As always I appreciate the help.
> My real question is this an interval then and will it be depreciated soon?

The Interval type won't be depreciated. Using to_char to convert intervals
to strings is being depreciated. This won;t cause a problem for extract
or similar functions.