[ntp:hackers] Leap second list file

Dave Hart davehart at gmail.com
Mon Jan 16 14:45:40 UTC 2012

On Mon, Jan 16, 2012 at 09:28, David J Taylor
<david-taylor at blueyonder.co.uk> wrote:
> Having said that, I can see why you might want more than one day's notice,
> so I would suggest making the notice period seven days, rather than one day,
> if that change can be accommodated.

Such a change would mean 7x more time to notice the upstream leap bits
using tools such as your leap second tracer.

> What would be the danger of firing the leap change seven times, though, were
> this done?  Is your "after the first hour" to avoid the chance of
> accidentally making more than one change per month?

No, as I said, it's to guard against misfiring a month early due to
small clock discrepancies.  The bits indicate only "at the end of the
month UTC insert/delete a leapsecond".  If I start exactly at the
beginning of the month and a dependent client's clock is a minute
slow, they might apply the leapsecond one month early.  By waiting an
hour, the likelihood of any client clock being far enough behind to
misinterpret the leap second's month is dramatically cut.  Waiting a
day would cut it more...

>  On reflection, perhaps
> the month is the more natural unit in any case.  Make it 24 hours so that
> any time zone doesn't get a potentially confusing indication, perhaps?

All times UTC, local time is not a factor here.

Dave Hart

More information about the hackers mailing list