[LEAPSECS] Leap Sec vs Y2K

[Warner Losh]

On 12/12/2010 18:11, Rob Seaman wrote:

> On Dec 12, 2010, at 2:53 PM, Poul-Henning Kamp wrote:

>

>> If Geophysicist announced leap seconds with at least 10 years

>> firm notice, 90% of the problems they cause would be eliminated

>> by the normal software update cycle.

> As you say, there are more than one possible way to redefine UTC - and in fact, what you describe would be perfectly conforming to the current standard.

Not quite.  If leap seconds were announced 10 years in advance, DUT1 
might exceed 1.0s if the prediction turns out to not be so good.  Of 
course, that would be better aligned than a DUT of 6s or 7s if leap 
seconds were completely eliminated.

It does seem like a reasonable compromise between different needs.  It 
would preserve the low-precision users (like PHP) to be more or less as 
accurate as they are today without modification, while not affecting the 
high precision users over much since they need DUT1 today anyway (the 
biggest effect would be to change the format DUT1 is published in).  
DUT1 broadcasts might also be a problem as well, but it is unclear who 
use those in preference to the internet values....

Warner