[LEAPSECS] Meeting with Wayne Whyte

Warner Losh imp at bsdimp.com
Tue Feb 1 12:55:22 EST 2011

On 01/31/2011 22:12, Mark Calabretta wrote:

> On Mon 2011/01/31 17:10:45 PDT, Warner Losh wrote

> in a message to: leapsecs at leapsecond.com

>

>> Earlier threads have called this the 'non-uniform-radix' problem. It

>> has been argued that there are no discontinuities in UTC, with the 59:60

>> notation offered as proof. However, this moves UTC from a uniform radix

>> that everybody is used to dealing with with to one with a

> We all deal every day with a non-uniform and variable radix counting

> system - "30 days hath September, ...".

>

> Leap seconds differ from leap days only in their unpredictability.

For dates in the past few hundred years (since the adoption of the
Gregorian calendar), you can 100% completely mechanically deduce the
right answer without the need for tables. You cannot do that for leap
seconds. If they were regularly scheduled, then that would be one
thing, but they are this random, drive-by event that you get 6 months
notice for.

>> This table-driven non-uniformity might or might not

>> technically be a discontinuity, but certainly is a pain in the back side.

> This is like saying that the Gregorian calendar might or might not

> technically be discontinuous. In truth it simply isn't discontinuous,

> there is no discontinuity on Feb/29 or any other day.

>

> As defined by TF.460, UTC is continuous, like the Gregorian calendar.

> That's all there is to it.

>

>> It is also a central problem of time_t: how do you map this

>> non-uniform-radix notation onto a uniform count that must always satisfy

>> properties that explicitly mandate a uniform-radix.

> Vide the mapping of calendar date to Julian Date.

True. However, this problem is also very mechanical and easy to do,
unless you also factor in which years which countries used what calendar.

> The fundamental problem is that there is no formula for determining

> when leap seconds occur.

Yes. And there cannot be such a formula. Unlike the period of the
orbit of earth, the rotation of earth is continually changing at a rate
that's unpredictable more than a few years out (apart from a general
trend, the slope of which scientists don't agree on (mostly because
different time periods have different slopes).

Warner