[time-nuts] 15 Seconds error...??

Attila Kinali attila at kinali.ch
Tue Jan 17 07:14:30 EST 2012


On Tue, 17 Jan 2012 06:57:22 -0500
Mike S <mikes at flatsurface.com> wrote:

> On 1/17/2012 2:04 AM, gary wrote:
> > My description of the problem would have been a thousand times better if
> > I mentioned monotonicity.
> 
> UTC is monotonic. The problem is with people who assume that minutes 
> always have 60 seconds. In UTC, they don't - they can have 59, 60, or 61 
> seconds.

Not 100% true. It depends on the representation of UTC.
If you represent it as a seconds counter (like unix time or similar),
then you can have seconds that are two seconds long and theoretically
even back steps.

In reality, you will have a monotonic increasing counter (but
not strictly monotonic!) and this is a problem. Hence POSIX introduced
strictly monotonic time source (but with an arbitrary starting point).

Be also aware that most computer systems or devices that need an
RTC are using a similar seconds counter. And all these systems
are affected by leapseconds if they are not defined to use an
offset value to calculate UTC.

			Attila Kinali

-- 
The trouble with you, Shev, is you don't say anything until you've saved
up a whole truckload of damned heavy brick arguments and then you dump
them all out and never look at the bleeding body mangled beneath the heap
		-- Tirin, The Dispossessed, U. Le Guin




More information about the time-nuts mailing list