Re: roundoff problem in time datatype

From: Dennis Bjorklund <db(at)zigo(dot)dhs(dot)org>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Dave Cramer <pg(at)fastcrypt(dot)com>, <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: roundoff problem in time datatype
Date: 2005-09-26 14:16:09
Message-ID: Pine.LNX.4.44.0509261552380.28238-100000@zigo.dhs.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers pgsql-patches

On Mon, 26 Sep 2005, Tom Lane wrote:

> > Actually, I think there is a case where 24:00 is a proper time. Isn't
> > it used for adding leap seconds ?
>
> No, I think the usual notation for a leap-second is '23:59:60'.
> We do allow 60 in the seconds field for this purpose.

Yes, and it can go up to 23:59:60.999999 (depending on how many fractional
seconds one want).

> I suppose there's another possible approach, which is to special-case
> the output of this value to look like '23:59:60' instead of '24:00:00'.

You would get the same problem with 23:59:60.9 which I guess you want to
round up.

One "solution" is to round '23:59:59.9'::time(0) up to '00:00:00'. That is
normally the next following time value after all. I know why you might not
want to round it "up" to 00:00:00, but it's one logical solution.

By the way, here is another example of the same problem:

# SELECT time '23:59:59.9' + interval '0.1';
?column?
----------
24:00:00

# SELECT time '23:59:59.9' + interval '0.11';
?column?
-------------
00:00:00.01
(1 rad)

--
/Dennis Björklund

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2005-09-26 14:28:31 Re: roundoff problem in time datatype
Previous Message Alvaro Herrera 2005-09-26 14:08:35 Re: "expected authentication request from server, but

Browse pgsql-patches by date

  From Date Subject
Next Message Tom Lane 2005-09-26 14:28:31 Re: roundoff problem in time datatype
Previous Message Tom Lane 2005-09-26 13:15:14 Re: roundoff problem in time datatype