[LEAPSECS] UTC going forward, one or two definitions? and what about NTP?

[mc235960]

Le 14 août 2013 à 09:59, Michael Spacefalcon a écrit :


> 

> 

> So why can't Dr. Gambis announce leap seconds by personal email as a

> private citizen, without any connection to any job?


  True I guess, but relying on one person to do the job is a bit much. In any case, the data needed to roll your own is readily available from the IERS. 
Who knows, there may be enough consensus among other camp followers here to chose the dates. 

 My own feeling is that there will be no change in the next half century or more.
 As Rob and some others here, I think that leap seconds are not broken. What is broken is the rigor applied to implementing them. 
It could be argued that it is theoretically impossible to guarantee correct implementation. The same argument would however be true for disseminating any other time scale.

It is another mater to deciding what a "civil" time scale should be, more of a  philosophical or even psychological one. 
  
  How long should a second be?
  Should seconds even be defined in it?  
  Should a minute have exactly 60 seconds?
  Should a day have exactly 86400 of those?
 etc.

 My own favorite, is to decouple civil time from the SI second and have it follow UT1 at much greater precision.  Say millisecond. 
This would mean implementation of rubbery civil seconds. But then , for the foreseeable future, a minute would always have 60 seconds, a day would have exactly 86400 seconds and common notions of the relation of time to the solar passage would be preserved.  
So for definitions, UTC would equal UT1 so no legislation needs changing.
Software that  can't handle leap seconds would be happy. 

As to implementation, no issue as leap seconds are fine for another 50 years or so. IUT-R would just need to give enough notice to ensure that everyone was ready.