[LEAPSECS] Unix Questions

[M. Warner Losh]

In message: <1281117013.26235.32.camel at localhost>
            Paul Sheer <p at 2038bug.com> writes:

: Can someone comment on this w.r.t. the protests about

: "Big-Ben time" eventually diverging from "sun-dial time"?


We have about 1 leap second every 18 months now, and that rate is
basically accelerating.  A leap hour would be needed in several
hundred years.


: Ie. is the ITU seriously considering such a divergence?


Yes.


:     Patrick Wallace (Rutherford Appleton Laboratory,

:     UK) suggested in his presentation that we should go

:     back to a 1960s style UTC that tracks UT1 more

:     closely and smoothly, an option that Dennis

:     McCarthy nicknamed UT1C in his overview of the

:     available options. 

: 

: Is UT1C still one of the options?


No.


: As a Unix developer I spend a lot of time fiddling

: with timestamps - I work with GSM/SS7 messaging which uses

: Unix 32-bit timestamps as an SMS text message identifier:

: 

: I would strongly prefer that Unix Time match UT1 identically.

: Of course this means that NTP services broadcast UT1

: instead of UTC and deprecate the leap second bit - which

: should work smoothly if they switch over on a day when

: UTC and UT1 exactly match - which happens every few years.

: 

: Are there any plans in this direction?


No.  Having Unix time track UT1 is problematic because nobody
broadcasts UT1.  Most of the ntp servers get their time from GPS,
which is based on UTC.

Also, UT1 has "rubber seconds" so you trade one set of problems for
another set of problems.

UTC is also the official time in many countries, so if NTP servers
were to broadcast UT1, then that would setup confusion on transactions
where <1s matters.  In today's trading environments, this can easily
happen.


: Is there any summary of the current proposals of the ITU

: and it's relation to and effect on Unix Time?

: 

: 

: Where can I find the definitive documents that the ITU

: is currently considering?

: 

: 

: When the ITU talks about changing UTC, what is their

: intention with respect to NTP services world-wide on

: which Unix Systems depend?


I think the plan is to ditch leap seconds, and leave the leap second
code that's in place and mostly kinda sorta works some of the time in
place to ever actually execute again.

Warner