[LEAPSECS] alternative to smearing

[Steve Summit]

John Sauter wrote:
> On Tue, 2017-01-03 at 13:28 -0500, Michael Rothwell wrote:
> > Why are you in such favor of leap seconds?
>
> I regard leap seconds as a reasonable compromise between the needs of
> civil time and of science.  Civil time needs a clock that tracks the
> days and the seasons.  Science requires a clock that measures time in
> precise intervals.  UTC provides both, using leap seconds to keep
> atomic time synchronized with the rotation of the Earth.

100% agree with John.  For thousands of years we've been doing a
better and better job of more perfectly aligning our chronometry
with the natural world.  If we abandon leap seconds, we're taking
a step (however small) backwards.

Another reason: I'm a C programmer.  The C standard says that
tm_sec has the range [0, 60].  But Posix won't let me see that 60.
I want to fix that.

Another reason: I'm a programmer at all.  One way of looking
at the present situation is that it's the physicists and the
astronomers who have done a better and better job of aligning our
chronometry with the natural world, but if we take this step
backwards, it's because it's too hard for the *programmers* to
properly implement the solution.  So there's definitely some
professional pride at stake here.

In fact -- for me, not talking about anyone else -- probably
the real reason I'm still tilting at this windmill is almost
perfectly summed up in that classic old screed about engineers:

	Ego

	The fastest way to get an engineer to solve a problem is
	to declare that the problem is unsolvable.  No engineer
	can walk away from an unsolvable problem until it's solved.
	No illness or distraction is sufficient to get the
	engineer off the case.  These types of challenges quickly
	become personal -- a battle between the engineer and the
	laws of nature.

But don't get me wrong; I do understand (the more so having
grappled with them up close and personal) that the obstacles to
"proper" handling of leap seconds are formidable.  I mentioned
being a programmer, but the other kind I am is a Unix programmer,
and it was Unix that gave rise to Richard Gabriel's essay about
the "New Jersey approach", which holds that simplicity of
implementation is sometimes even more important than strict
correctness.  And it only recently occurred to me that there's
no better example of this tradeoff than deciding that
monotonic-seconds-since-1970 is "good enough", is practically
better even though it's also "worse" in not having any good way
of encoding leap seconds.

Steve Summit