[LEAPSECS] The Debate over UTC and Leap Seconds

[Poul-Henning Kamp]

In message <20100810194307.GB4307 at cox.net>, Greg Hennessy writes:

>On Tue, Aug 10, 2010 at 11:14:15AM -0600, M. Warner Losh wrote:

>> I think that he means that the WP7A folks are telling the software

>> community that either they suck, or it really isn't a problem.  

>

>Well, they may have no desire to tell the software community they

>suck, but the software certainly sucks, in the sense that it doesn't

>represent reality, i.e. leap seconds.


Ohh, absolutely:  95% of all programmers thing they are above average
and in reality 99% of them suck at what they do, and experience shows
that there is not a shit we can do about it.


But the real problem here is one of tacit assumptions that get
invalidated along the line, starting with the chosen 1850-ish LOD
value over the "nobody will care but us scientists" from where
it switches into the software domain with "Who cares ?  Leap-seconds
are hard, lets ship a product."

I don't think anybody has any moral right to throw stones here, but
the prevailing attitude is one of "If everybody just do what I tell
them to, the world will be saved" and that is not conductive to
cooperation.

I have a high regard for astronomers, almost became one myself[1],
but I have to say that the leap-second debate has given me
the impression that their ability to see outside their own
little world is insufficient for this purpose.  The paper that
prompted this discussion being a great example thereof.

It must be nice to be in an environment where everybody has a phd,
or engineering degree, but the mental model from that environment
clearly does not transfer.

Out here in the rest of the world, we have former tram-drivers
programming computers and college dropouts writing the standards
and programming languages they use for it[2].

Just saying "They should just do it right" is simply not an option
that can fit inside any educational programmes ambitions and budget,
and Dijskstra curse applies 100%[3]

We _have_ to arrange it so the lousy programmers must make an effort
to get it _wrong_, the other way simply do not work outside astronomers
fantasies.


>The two solutions are: 1) change the software standards to match

>reality, or 2) change reality to match the software standards. If

>there was a third, I've missed it.


There is a 3rd but nobody seems willing to take it seriously:

	"Make leap seconds at lot less painfull by announcing them
	10 years in advance".   That makes them "calendar" instead
	of "timekeeping" for the software people, and moves the
	update problem from real-time networking requirements to a
	software distribution issue.  20 years would be better, but
	10 years is livable.

As a timegeek I *like* leap-seconds, they're the unmarked 1-penny
black for people like me.

But as a computer professional, I must also insist that planes do
not start dropping out of the sky because of them, given the
level of programmers we have access to.

Poul-Henning


[1] Until I realized that making a living in astronomy involved
waiting for some 70+ year old geezer, who got plenty of exercise in
pristine mountain-top-air, falling off a cliff, before an opening
would appear, because they sure didn't seem to die from old age.
I'm not kidding: at my first astronomy conference a 73 year old
bloke started out talking about this 15 year observation programme
he just started...

[2] To see what we are up against, search around a bit for date
formats in spreadsheets, leap-years are still a challenge there.

[3] "I have only a small brain, and must live with it" [while
computers gets bigger and faster].  He was complaining about
the first computer with 64k storage.

-- 
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.