Re: [LEAPSECS] PTTI leap second discussion

From: Steve Allen <sla_at_UCOLICK.ORG>
Date: Sun Apr 1 00:39:31 2001

On Fri 2001-03-30T15:10:46 -0500, Demetrios Matsakis hath writ:
> The following is to be published in the Proceedings of PTTI-2000.

Many thanks for providing an advance copy.

Steven Malys of NIMA reports that many long-running, large operational
systems of DoD and other governmental agencies have no problems with
leap seconds. They are apparently just part of the routine operation,
and the cessation of leap seconds would cost money. He nicely
explains the nature of the costs associated with change to the status

There seems to be broad agreement that GPS -- a system with hardware
designs implemented over 20 years ago -- is time with leap seconds
done right (so long, points out Demetrios Matsakis, as people remember
about the 19 second difference between the GPS epoch and TAI (or 13
seconds between current UTC and GPS)).

Judah Levine reports on the wonderful concept of some kind of NTP
service which permits the user to choose TAI or UTC. (If that were to
come along with estimates of UT1, astronomers could ask for no more.)

Thomas Celano works for a timing hardware company founded in 1990 --
some 18 years after leap seconds were introduced -- yet he speaks of
his customers shutting down whenever a leap second happens. On the
other hand, he states that predictable leap seconds would reduce the
costs of handling them.

It would be very enlightening to know why systems implemented with
hardware 10 years more advanced than GPS cannot handle, or have
significant costs handling, leap seconds. In particular it would be
interesting to know what he means by "predictability" and what levels
of cost ameliorations are provided by different amounts of

Robert Nelson argues that the sun is irrelevant to modern civilization,
and also that the conversion of UTC to another offset flavor of TAI
along with the creation of a new name for what we now call UTC is
preferable to taking a one-time, 32-second leap. Unfortunately the
result of this would be just one more kind of offset time with the same
sort of problems as the 19 (or currently 13) second GPS time difference
mentioned by Matsakis.

Obviously neither of the two opposing camps wants to undertake the
detailed tally of the actual costs of change or not-change. But at
least from Malys we have an explanation of the nature of the costs
feared by the "don't change it" camp, whereas from Celano and the "do
change it" camp we have no explanation of the nature of the costs. A
good explanation (even without a tally) of the ongoing costs incurred
by systems which do not tolerate leap seconds owould help a great deal
to further mutual understanding between the two camps of participants.

Steve Allen          UCO/Lick Observatory       Santa Cruz, CA 95064      Voice: +1 831 459 3046
PGP: 1024/E46978C5   F6 78 D1 10 62 94 8F 2E    49 89 0E FE 26 B4 14 93
Received on Sun Apr 01 2001 - 00:39:31 PST

This archive was generated by hypermail 2.3.0 : Sat Sep 04 2010 - 09:44:54 PDT