[LEAPSECS] UTC fails

[Brooks Harris]

On 2015-03-12 11:57 AM, Stephen Colebourne wrote:
> On 12 March 2015 at 05:21, Steve Allen <sla at ucolick.org> wrote:
>> On Wed 2015-03-11T11:04:57 -0700, Tom Van Baak hath writ:
>>> The entire purpose of UTC is to provide a single timescale for all
>>> human-related activity.
>> And UTC has failed miserably.  POSIX says UTC has no leaps.
>> Google says UTC has occasional days with stretches of seconds which
>> are of varying lengths.  De facto, there is no single UTC time scale.
> But what so many miss is that what is needed to fix the problem is very small.
>
> 1) Reliably send leap second data out to the world (recently discussed
> here and at tz-dist)
Yes. That's the crucial missing link.
> 2) Announce leap seconds a bit further in advance or on a regular schedule
Yes.
>
> 3) Define a time-scale, UT-86400, that roughly follows UTC but always
> has 86400 "second-like" subdivisions (as per the Java time-scale)
That's similar to NTP and POSIX. These timescales work just fine for 
creating "broken down time" except for the 23:59:60th count (or rollover 
at 23:59:58) and the fact their absolute seconds offset (time_t) does 
not include the Leap Seconds.

>
> 4) Provide one or more *agreed* and *standardised* mechanisms to map
> UTC to UT-86400 (eg. UTC-SLS and Google smear)
Yes, but not to non-deterministic work-around things like Google Smear!
>
> The fact that we don't have a name or agreed standard for the thing
> that most people want (outside the time-nerd community) is very sad.
> UT-86400 is a working name, I'm sure someone can think of a better
> one.
Yes, in a way. The mismatch between UTC and the many timescales with 
86400 second days one part of the the difficulties. There are many 
"86400 second day" timescales that are not exactly the same (NTP and 
POSIX, for examples) so there's already many potential names. We do have 
the Gregorian calendar timescale, but this can't deal with Leap Seconds 
by itself is related to each of the timescales in different ways. These 
timescales exist and are in wide use so they can't be pulled back. 
Gregorian, TAI, and UTC are the closest things to "common" you're going 
to get.
>
> The work needed isn't hard. I just wish that rather than destroying a
> sensible solution to keep us in line with solar days, effort would be
> put into defining the above.
Conceptually not difficult depending on who you're talking to, but 
arriving at consensus for standardization is a whole other matter. It 
can be done, it needs to be done, but it won't be easy.

-Brooks

>
> Stephen
> _______________________________________________
> LEAPSECS mailing list
> LEAPSECS at leapsecond.com
> https://pairlist6.pair.net/mailman/listinfo/leapsecs
>
>