[LEAPSECS] The relation between calendars and leap seconds.

[Poul-Henning Kamp]

In message <1036CB00-CB83-4F7F-B2CB-C1CF78DB6CEA at noao.edu>, Rob Seaman writes:



>Continue compiling a coherent, complete and self-consistent set of  

>requirements and we can get around to testing the various options.



The most important requirement is obviously that the proposed
change, can possibly be made to happen.

Using the most lenient standard of review we can, we assume that a
"scientifically perfect" solution is found.

By "scienfically perfect" we obviously do not assume that "we will
stabilize the Earths rotation", but a solution that works inside
the bounds of our geophysical knowledge and models of the physical
reality.

Likewise, the solution is not a technological magicians trick,
such as "perfect atomic clocks surgically connected to our brains
right after birth" or "all devices must be on the internet" etc.

In other words, we assume a solution that, through one or more time
scales, caters to the various needs of humanity, along with such
rules and observational frameworks that will be required for the
conversion between these time scales.

We totally disregard small details such as naming, chosen tolerances,
and thresholds, assuming that the best possible choices have been
in all such matters.


Under this hypothetical scenario, please flesh out this sequence
of events:

    A.  Rob Seaman announces the scientifically perfect NewTime to
        the world.

    [ Please fill in what happens here ]

    H.  Scientists, all over, generally applaud this as sensible.
	(With the usual small, but loud, fraction of emeriti who
	grumble on general principles of age, and are revered but
	not necessarily listened to, for the very same reason.)

    [ Please fill in what happens here ]

    N.  NewTime becomes the legal basis of civil timekeeping in all
	of USA & EU.
	[please insert realistic year this could happen]

    [ Please fill in what happens here ]

    Z.  NewTime is implemented on 99% of all USA & EU based IT
        systems which react according to what time it is.
	[please insert realistic year this could happen]


Having done that, please address the probability that the entire
plan will be derailed by:

    ?.  Computer consultants estimate the cost of implementation
	of NewTime in legacy systems to $BIGNUM, and offer the
	alternative of just dropping leapseconds from UTC as a
	virtually no-cost change.

This event which can be assumed to happen right after A and
certainly long before N.


Poul-Henning

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.