In message 90362.1122573870@phk.freebsd.dk, "Poul-Henning Kamp" writes:

                        ^
                       not

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

At 01:58 PM 7/28/2005, Bjorn Gabrielsson wrote...

It will break if the system is built on the basis that UTC is within 0.9 seconds of UT1, which is how UTC is currently defined. The legal system in the US (and many other countries) is based on solar time, so it would break legal timekeeping. There may be systems dealing with satellite tracking/orbit maintenance which might break. There are astronomical systems which would break. I don't know what else, but the point is, neither do the people proposing to break UTC with short warning and little consultation.

It is only kinda based on solar time.  And only at certain locations,

It is true that the definition of the time zones, in 15 USC 261
states, in part:

the standard time of the first zone shall be based on the mean
solar time of the sixtieth degree of longitude west from
Greenwich... (and so on for each of the time zones)

"based on" can mean a lot of different things.  15 USC 260a
explicitly says that this time trumps any laws that the states may
have on an official time (later sections allow certain states to opt
in/out of daylight savings time, and define certain time zones).

However, there's some weasil words in the next section, 16 USC 262,
which seem to indciate that the actual legal time is determined by the
govnerment:

Within the respective zones created under the authority of sections
261 to 264 of this title the standard time of the zone shall insofar
as practicable (as determined by the Secretary of Transportation)
govern the movement of all common carriers engaged in interstate or
foreign commerce. In all statutes, orders, rules, and regulations
relating to the time of performance of any act by any officer or
department of the United States, whether in the legislative,
executive, or judicial branches of the Government, or relating to the
time within which any rights shall accrue or determine, or within
which any act shall or shall not be performed by any person subject to
the jurisdiction of the United States, it shall be understood and
intended that the time shall insofar as practicable (as determined by
the Secretary of Transportation) be the United States standard time of
the zone within which the act is to be performed.

The operative phrase being "as destermined by the Secretary of
Transportation".  It is up to this department of government to
delegate the definition to someone.  The time and frequency division
of NIST have this delegation, and provide the official time.  If they
say that the time is a little fast or a little slow, it is still based
on the solar mean time, but it isn't the solar mean time exactly.  The
US already doesn't observe exactly the solar mean time, but an
approximation of it based on UTC, as recovered by NIST.

I have not been able to find the actual regulations delegating this to
NIST (only statements to that effect on the NIST web site), so I'm
unable to tell how much deviation is allowed from the mean solar time
to still meet the statuatory langauge of 'based on'.  Clearly this can
be as large as 0.9s, since that's the standard that's been adopted
today.  But who is to say that it can't be more than that and still be
based on the mean time?  Without looking at the actual current
regulations that have been promulgated, it is hard to say, with
certainty, that eliminating leap seconds would break legal time.  I
tried hard to find them, but alas couldn't.

BTW, the 15 USC doesn't say anything at all about leapseconds or leap
seconds.  Maybe the politicians are the smart ones here :-)

Warner

mikes@flatsurface.com (Mike S) writes:

Seems to be a minor concern.

Legal time is often based on UTC. A slight adjustment of the
definition of UTC will not break legal timekeeping.

Thats a gliding scale... if system A breaks at DUT1 of 1.0seconds
B & C might die at 5 and 10 seconds, giving plenty of time to patch
the systems. System A does seem to live on the edge already, maybe a
good candidate to patch earlier...

Leapseconds are known with short warning... Maybe the astronomers have
not been following the lengthy discussions among the real timekeepers of
today. Whos fault is that?

"Break UTC" - or simplify/improve UTC.

--
Björn

mikes@flatsurface.com (Mike S) writes:

If specifying "based on mean solar time" does the US (or other less
important parts of our globe) specify a reference to which particular
epoch it should be based on? -- a careful choice would ofcause
eliminate leap seconds...

--
Björn

At 06:50 PM 7/28/2005, Bjorn Gabrielsson wrote...

Personal opinions don't count. To you it's minor, to someone else major. What's your point?

And often not. What's your point?

You are able to assuredly identify which systems will fail, and when? It must be nice to have omniscience.

Leap seconds are known with about 6 months warning. UTC has been well defined for over 30 years. Like it or not, "time" is astronomical, both historically and in actual civil use. ITYM "self-proclaimed timekeepers."

It is up to the organization tasked with a duty to carry out that task responsibly. Failing to notify/seek input from obviously affected parties is irresponsible.

How would UTC be improved? It would be exactly parallel with TAI, which is already available for use.

defined for over 30 years. Like it or not, "time" is astronomical, both
historically and in actual civil use. ITYM "self-proclaimed timekeepers."

Correct, and that's why, in the very long term,
TAI is not suitable to civil time. And that's also
why UTC, whether implemented with leap
milliseconds, leap seconds, leap minutes,
or leap hours, is suitable to civil time. The key
thing about UTC is the C, the coordination
between atomic and astronomical time. The
UTC leap hour proposal honors that.

/tvb

At 07:28 PM 7/28/2005, Tom Van Baak wrote...

The key to your claim, "coordinated," does not refer to the synchronization of UTC with UT1. It refers to coordination of multiple physical clocks. As David W. Allan states: "Time and frequency coordination is the process of combining primary standards to generate coordinated T/F standards for the world or for a particular country." - http://tf.nist.gov/timefreq/general/pdf/619.pdf Here's another NBS document specifically addressing the coordination of UTC, not with UT1, but between clocks: http://tf.nist.gov/timefreq/general/pdf/1744.pdf

The "UT" in UTC is of course in reference to Universal Time, and it's variants, all directly linked to earth rotational time. UTC was specifically created to  offer a close approximation of UT1, with the precision of coordinated time (TAI).

Removing leap seconds makes it indistinguishable from TAI with the exception of a fixed offset, and the ability of lazy sysadmins to claim they're keeping time correctly. Such a time scale can not correctly be called "UT," as it would no longer a variant of Universal Time.

The promise of a "leap hour" millennia in the future is disingenuous lip service. If that time should come, there will once again be ineffectual sysadmins complaining that it's too hard and too expensive.

synchronization of UTC with UT1. It refers to coordination of multiple
physical clocks

Mike,

Well, there's a lot of coordination going on.

TAI is the coordination of multiple, physical
atomic clocks.

UTC is the coordination of astronomical (UT1)
and atomic (TAI) timescales.

/tvb