[LEAPSECS] Daylight Time snafus

[Brooks Harris]

On 2014-02-24 03:25 PM, Matsakis, Demetrios wrote:

> Sometimes it doesn't matter how much notice is given.  This morning I got an email from someone asking why his clock just jumped to daylight time.    I've had many emails from people whose clocks incorporate a change late, which could be due to radio reception problems among other things.   I've also gotten some from people whose clocks are pre-programmed with the old American daylight time rules, or European ones.   But this is a first for early daylight.

>

> As always programming errors are the default guess.   Last leap year we even suspected one product had an issue related to February 29.  Only the manufacturers know for sure.

I'm afraid, and I'd bet, the manufacturers don't know for sure either.

In an environment where the underlying timekeeping standards are complex 
and fractured, and where the underlying computer standards, mostly ISO c 
and POSIX, have known inadequacies, its little wonder there are 
interoperability problems if attempting to represent accurate civil time 
(whatever that might be :-| ). Implementers are faced with extensive 
research and, when that's done, left with many unknowns. There's no 
reference implementation for guidance and no reasonable testing procedure.

Windows side-steps the problem by simply waiting for, and relying on, 
synchronization to NTP severs at the cost of being wrong until the 
reset. Unix/Linux attempts to maintain the time in the kernel making it 
vulnerable to failure, especially if its based on the flawed POSIX 
specs, not to mention difficulties with tz database. Meantime there is 
no standardized automated way to obtain the required metadata.

Its not a matter of UTC with Leap Seconds being complicated - the theory 
and basic implementation by the BIPM and IERS is sound and a remarkable 
achievement of civilization. The problems lie in digital implementation 
and deployment - society's current challenge generally, not only in 
timekeeping. We should not abandon 3000 years of timekeeping technology 
because some systems cause inconvenience and we are too lazy to refine 
flawed standards.

-Brooks





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://six.pairlist.net/pipermail/leapsecs/attachments/20140224/3b31c1f1/attachment.htm>