From: Nem Tudom <ellenallhatatlan@gmail.com>
Message-ID: <c3520c84-33d3-47f2-a859-091f369bdc91@gmail.com>
Date: Mon, 27 Jan 2025 21:01:59 +0000
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Content-Language: en-US
To: pgsql-general <pgsql-general@postgresql.org>
Subject: Issues with EPOCH-s, TIMESTAMP(TZ)-s and leap seconds.
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: <https://www.postgresql.org/message-id/c3520c84-33d3-47f2-a859-091f369bdc91%40gmail.com>
Precedence: bulk


Hi all,


I'm having trouble understanding matters related to TIMESTAMP(TZ)-s and 
leap seconds - my machine runs on UTC so as to remove any issues related 
to the zones.

 From here: https://en.wikipedia.org/wiki/Leap_second,

There have been 27 leap seconds added to UTC since 1972.


But, when I run this fiddle (see bottom of this email link)

https://dbfiddle.uk/wxvmzfJb

(first snippet - 2015 -> 2016) I get a "nice" even number for the EPOCH 
of, 00:00:00 2016 , say (= 1451606400) - now, with 27 leap seconds since 
1972, I would expect that number to be (something like) 1451606427?

I thought that the EPOCH was the number of seconds since 1970-01-01 
00:00:00? Is this incorrect?

Also, (first snippet again), why is the TIMESTAMPTZ 23:59:60 2015 even 
allowed?

Now, we come to the second snippet (2016 -> 2017), I get *_exactly_* the 
same behaviour!

I was expecting to see that '2016-12-31 23:59:60'::TIMESTAMPTZ would 
work (leap second) and then that '2017-01-01 00:00:00'::TIMESTAMPTZ 
would have incremented by 1 second?

I'm puzzled. Does PostgreSQL take leap seconds into account? Does anyone?

Any help, advice, recommendations, URL-s, references &c. appreciated.


E...