Re: [LEAPSECS] more media coverage

From: Rob Seaman <seaman_at_noao.edu>
Date: Fri, 11 Jul 2003 12:11:10 -0700 (MST)

Ed Davies says:

> My 757/767 captain friend couldn't think of any actual Earth
> rotation dependencies -

I commend you for continuing to press this discussion with folks
outside the leap second loop. I suspect that many of the folks on
this list, like I, were active in their organizations Y2K remediation
efforts. The expense of Y2K was no only - or perhaps even mostly -
contained in fixing bugs and testing the fixes. A large fraction of
the expense was in the time and money needed to inventory every system
that could even potentially require Y2K related changes.

We have no business discussing a fundamental change to international
time standards without performing a similar inventory. Not only is
such an inventory needed before implementing a change - such an
inventory is needed before designing the change.

Before Y2K, highly competent members of our staff (equivalent in
experience to your 7N7 captain) opined that there were likely no major
Y2K remediation efforts that the observatory would have to undertake.
It is indeed true that the Y2K inventory that I supervised on our major
image processing package (used by most of the world's astronomers) only
turned up a couple of dozen routines out of 10,000 or so that required
changes. But those were key routines for many activities and an easy
test (using the computer I'm currently typing this on) showed that our
unremediated software was going to fail completely. Those two dozen
routines required relatively simple bug fixes. There was also a new
interface that we found prudent to write that now isolates all date
conversions within the system.

For this system, the characterization of the problem was by far the
most expensive step. The expense of validating the correctness of
the fixes was not far behind. Actually implementing the fixes was
relatively easy - but note that Y2K really involved only bug fixes
and very little system redesign.

Another expense of Y2K was the revision of dependent standards. Several
of the members of the astronomical software community who also subscribe
to this list were active in the revision of the astronomical FITS data
interchange standard to support Y2K related changes. This involved
large numbers of individuals from many organizations in our community
who each spent many weeks or months on this issue. The details were
discussed for three or four years prior to 2000-1-1. Our efforts have
yet to completely cease.

And yet another expense of Y2K to the astronomical community was what
I would say falls into the category of random surprising absurdities
of legacy systems. The observatory scheduled a variety of activities
to characterize the Y2K behavior of expensive, one-of-a-kind systems.
The more recent systems were relatively well behaved and well understood.
Typically the folks who designed those systems were still on staff to
make the necessary fixes. But astronomy is not unique on relying on
a wide range of legacy systems, some of whom are poorly documented and
most of whom were long since orphaned by retired or departed staff.

In short, two of our oldest telescopes were controlled by PDP's running
forth (one of these still is) - and we discovered that those two
telescopes were going to track the sky backwards after the millennium.
Were they fixed in time? Yes. Was it easy? Relatively - if you ignore
the fact that some very busy staff member had to relearn obsolete
technologies purely for this purpose.

It may well be that all of the world's interlocking technical, social,
economic, commercial, religious, political, computational, and scientific
systems are completely uncaring about civil time continuing to approximate
time-of-day. It may well be that only a few obscure loonies like
astronomers will have to change anything at all - and that the lives
of millions of hardware and software engineers will be made easier
for generations to come.

It may be that there are NO other international standards that depend
on the current assumptions of the UTC standard. It may be that there
are NO de jure or de facto (UTC ~ GMT) dependencies in any jurisdiction
on the planet. It may be that not a single religious sect anywhere on
the globe will care about the secularization (pun intended) of the
world's clocks. It may be that there will be not a single negative
economic consequence. It may be that (other than astronomers and
traditional sectant navigators) not a single operational procedure or
technical software/hardware system will have to be changed. It may
be that only a few publishers will have to update tables in almanacs
and that they will view this as a way to sell more books. It may be
that no legacy systems will have to be updated or retired. It may be
that the world's programmers will either be able to ignore this change -
or will view it as another full employment opportunity like Y2K.

But shouldn't we fully inventory and characterize and test the various
such named and unnamed systems BEFORE we decide to make a change?

Rob Seaman
National Optical Astronomy Observatory
Received on Fri Jul 11 2003 - 12:11:24 PDT