Date: Mon, 11 Oct 2021 17:56:12 -0400
From: Bruce Momjian <bruce@momjian.us>
To: "David G. Johnston" <david.g.johnston@gmail.com>
Cc: Zhihong Zhang <zhihong@gmail.com>,
	pgsql-bugs <pgsql-bugs@lists.postgresql.org>
Subject: Re: Epoch from age is incorrect
Message-ID: <20211011215612.GE643@momjian.us>
References: <E756CE54-1946-4DC8-9F63-9BB54A07EF63@gmail.com>
 <CAKFQuwb5tRgZs0K_2iAzYdQ3uYcxMJ+zLObpYmx05HDpSiLCiQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: 
 <CAKFQuwb5tRgZs0K_2iAzYdQ3uYcxMJ+zLObpYmx05HDpSiLCiQ@mail.gmail.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Archived-At: 
 <https://www.postgresql.org/message-id/20211011215612.GE643%40momjian.us>
Precedence: bulk

On Mon, Oct 11, 2021 at 02:52:59PM -0700, David G. Johnston wrote:
> On Monday, October 11, 2021, Zhihong Zhang <zhihong@gmail.com> wrote:
> 
> 
>     They are several days apart. Maybe leap years are not counted properly?
> 
> 
> That is related.  Intervals supply integer years, months, and days (for
> example).  Given an integer month (or year) value there is no way to accurately
> compute the number of days involved (or any subfield of day) since you lack
> knowledge of which months (years) they are (generally).  So extracting an epoch
> from an interval is an inherently unwise thing to do.  Having done it,
> comparing its result to a timestamp epoch computation is meaningless.  If you
> can do the later then just do it as it will provide the expected answer while
> the interval epoch is inherently problematic.

Yes.  Also, this blog has more details:

	https://momjian.us/main/blogs/pgblog/2020.html#August_3_2020

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  If only the physical world exists, free will is an illusion.