[LEAPSECS] An example

[Poul-Henning Kamp]

In message <536635d43949d1db19f5fa0ce101632e.squirrel at mx.pipe.nl>, "Nero Imhard
" writes:

>Op 02-11-10 23:56, Poul-Henning Kamp schreef:

>

>> There very much is: With each passing day, the probability that the next

>> leap second will kill somebody because of sloppy engineering increases

>> by a possibly non-trivial amount.

>

>Then why aren't you demanding proper engineering instead? Giving in to

>sloppy engineering won't be any improvement, now would it? In the long run

>it will be even more dangerous.


Proper engineering is all about picking your challenges in light of
their benefits.

Leap seconds carry no discernible benefit to non-astronomers so
even if they were trivial to code correctly, they would not be
worth it.

The only plausible way to handle them in a more-likely-correct-than-wrong
way, would be if they were announced at least 10 years in advance, so
they could be diseeminated with software updates at the OS level.

But in the absense of any tangible benefits, it is much simpler and
cheaper to just stop calling them, which makes all POSIX software
correct, by definition, even if it has been coded to deal with
leap seconds.

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.