[time-nuts] Leap second glitches on NTP using Z3801A

[Tom Van Baak]

> Ascending from Lurk Mode, I have a (possibly stupid) question: according to
> http://tycho.usno.navy.mil/leapsec.html
> and Tony Jones's book "The Story of Atomic Time" GPS time does not
> account for leap seconds, So why does it alert you to them?

Eric,

"Does not account for" maybe isn't quite what happens, but
right, GPS time rolls along continuously without regard to
variations in astronomical time.

A leap second adjustment count is included in the GPS data
stream so that receivers can output true UTC time stamps, if
that conversion is required.

Some level of advanced warning of a leap second is necessary
so that when the moment arrives the receiver can either delete
the UTC second called 23:59:59 (negative leap second) or insert
an extra UTC second called 23:59:60 (positive leap second).

Theoretically, a GPS receiver might only need one second of
advanced warning to do the right thing in the last UTC second
of the last day of the month, but for a variety of good reasons,
a couple of days or even a couple of months of advanced
notice is better for everyone involved.

/tvb