[LEAPSECS] Leap seconds decision deferred until 2015

[Warner Losh]

On Jan 19, 2012, at 4:32 PM, Stephen Colebourne wrote:

> So, why not explore it. Properly. On this list.


Rob Seaman doesn't like it when I do that.  He always tries to draw the discussion back to the ITU proposal, even when I'm trying to find middle ground.  Maybe now that it has been shelved, we can make some headway.


> This list has key knowledge of the issues, yet has failed to focus that into specific alternate proposals that I have seen. I would encourage list members to work out some more formal alternatives and post them in a suitable thread. It would be far better for the next working group to be able to be sent 2 or 3 more structured options between "do nothing" and "kill entirely".


I'm surprised that nobody had the 'announce earlier' or 'phase in DUT1 > 1s over decades' card out of their pocket.  At this point it is should be advised that it is unsafe to assume that DUT1 < 1s for new applications and operators should not assume that DUT1 < 1s and those applications that care should obtain it from the internet.  Publish it via a http port and then anybody that's on the net can get it.  At least it would start to get rid of the deeply seeded assumption in case it becomes too expensive for others to carry the burden of DUT1 < 1s.  If it could go to 10s, then we could announce leaps 5years in advance, if not 10, for example.  It wouldn't drift unbounded and would still be a universal time, just one with slightly larger, but more available, error bars.

Warner