Date: Sat, 1 Sep 2018 19:37:36 -0400
From: Bruce Momjian <bruce@momjian.us>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: PostgreSQL-documentation <pgsql-docs@postgresql.org>
Subject: Re: AT TIME ZONE correction
Message-ID: <20180901233736.GA25700@momjian.us>
References: <20180901225557.GA22406@momjian.us>
 <24277.1535844643@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <24277.1535844643@sss.pgh.pa.us>
User-Agent: Mutt/1.5.23 (2014-03-12)
Precedence: bulk

On Sat, Sep  1, 2018 at 07:30:43PM -0400, Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > Looking over the AT TIME ZONE docs, I think they are subtly confusing. 
> > The order of conversion specific in the first example should _start_
> > with the assumption of local time zone for the time stamp, not something
> > that happens after AT TIME ZONE is applied.  The ordering in current
> > docs makes the second example confusing too.
> 
> > The attached patch fixes this.
> 
> I think it's you that are confused.  The text as written is correct,
> or at least arguably so; your revision is definitely incorrect.
> 
> SELECT TIMESTAMP '2001-02-16 20:38:40' AT TIME ZONE 'MST';
> 
> The way I'd describe this is that we start with a timestamp,
> which has no particular timezone.  The AT TIME ZONE says "Assume that
> this zoneless timestamp is in MST, and convert it to timestamp with
> time zone (which will be in UTC, internally)".  Then after that, the
> UTC timestamptz value is converted to PST8PDT for display purposes,
> but that's done by timestamptz_out not AT TIME ZONE.
> 
> If we were going from PST to MST as your patch describes it, the
> output would be one hour later not one hour earlier than the input.
> 
> SELECT TIMESTAMP WITH TIME ZONE '2001-02-16 20:38:40-05' AT TIME ZONE 'MST';
> 
> Here we've got a time value that was initially given in EST (-05),
> but was converted to UTC by timestampz_in.  Then the AT TIME ZONE
> says "Please convert this UTC value to MST, and emit it as a zoneless
> timestamp" (which will not be subject to any further conversion when
> it's displayed).
> 
> The existing text is indeed a bit deficient, because it fails to
> draw a clear boundary between what the AT TIME ZONE operator is
> doing and what is being done by the timestamp(tz) I/O functions.
> But you're not making it better.

Yes, I am still researching and realize my diff is wrong.  Let me keep
working and I will repost.

-- 
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

+ As you are, so once was I.  As I am, so you will be. +
+                      Ancient Roman grave inscription +