From sla at ucolick.org  Fri Aug 15 02:09:52 2008
From: sla at ucolick.org (Steve Allen)
Date: Thu, 14 Aug 2008 23:09:52 -0700
Subject: [LEAPSECS] choose the form
Message-ID: <20080815060952.GA23990@ucolick.org>

It's getting close to the time when WP7A will be deciding on the
fate of leap seconds.

I am willing to agree that the underlying time scales of broadcast
signals and computer systems need to be free of leap seconds.

I am not willing to agree that civil time can drift from the sun.

I also maintain that the "right" forms of the zoneinfo files already
are proof of concept.  They demonstrate that computers can have an
uniform underlying time scale while still providing a civil time scale
that has leap seconds.  The technical issues are basically solved.

Inspired by Seidelmann and Fukushima's pictorial form, I have been
working on these
http://www.ucolick.org/~sla/leapsecs/deltat.html
I'm not done with details and decorations, but the basic ideas are
there now.

What remains is for the ITU-R to choose the form of the future.

--
Steve Allen                 <sla at ucolick.org>                WGS-84 (GPS)
UCO/Lick Observatory        Natural Sciences II, Room 165    Lat  +36.99855
University of California    Voice: +1 831 459 3046           Lng -122.06015
Santa Cruz, CA 95064        http://www.ucolick.org/~sla/     Hgt +250 m

From phk at phk.freebsd.dk  Fri Aug 15 04:56:55 2008
From: phk at phk.freebsd.dk (Poul-Henning Kamp)
Date: Fri, 15 Aug 2008 08:56:55 +0000
Subject: [LEAPSECS] choose the form
In-Reply-To: Your message of "Thu, 14 Aug 2008 23:09:52 MST."
	<20080815060952.GA23990@ucolick.org> 
Message-ID: <2173.1218790615@critter.freebsd.dk>

In message <20080815060952.GA23990 at ucolick.org>, Steve Allen writes:

>I am willing to agree that the underlying time scales of broadcast
>signals and computer systems need to be free of leap seconds.

Thanks, a much needed conversion :-)

>I am not willing to agree that civil time can drift from the sun.

I doubt they will, but fiddling with it in hour-size bits every
few hundred years is much more likely to work than one second bits
ever other year, so that would be my preference.


>I also maintain that the "right" forms of the zoneinfo files already
>are proof of concept.  They demonstrate that computers can have an
uniform underlying time scale while still providing a civil time scale
>that has leap seconds.  The technical issues are basically solved.

Well, you havn't by any stretch of the imagination solved the problem
of how to get the zoneinfo files onto the computers.

Just this last week, a DG Ecclipse/8000 system was powerd off, it
has been doing ATC radar stiching for the last 20 years or so.

Do I need to tell you when they last got a software patch for their
operating system ? :-)

>Inspired by Seidelmann and Fukushima's pictorial form, I have been
>working on these
>http://www.ucolick.org/~sla/leapsecs/deltat.html
>I'm not done with details and decorations, but the basic ideas are
>there now.

Cool work, do I spot another disciple of the cult of Tufte ? (otherwise
buy his books (www.tufte.com), you will love them :-)

Poul-Henning

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at 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.

From tvb at LeapSecond.com  Fri Aug 15 06:06:33 2008
From: tvb at LeapSecond.com (Tom Van Baak)
Date: Fri, 15 Aug 2008 03:06:33 -0700
Subject: [LEAPSECS] choose the form
References: <20080815060952.GA23990@ucolick.org>
Message-ID: <04d601c8febe$98f40770$0300a8c0@pc52>

> I also maintain that the "right" forms of the zoneinfo files already
> are proof of concept.  They demonstrate that computers can have an
> uniform underlying time scale while still providing a civil time scale
> that has leap seconds.  The technical issues are basically solved.

I know I've been quiet on your zoneinfo thread last year, but
my question is what percent of equipment or devices affected
by leap seconds use zoneinfo files? 10%? 1%?

/tvb


From phk at phk.freebsd.dk  Fri Aug 15 06:10:01 2008
From: phk at phk.freebsd.dk (Poul-Henning Kamp)
Date: Fri, 15 Aug 2008 10:10:01 +0000
Subject: [LEAPSECS] choose the form
In-Reply-To: Your message of "Fri, 15 Aug 2008 03:06:33 MST."
	<04d601c8febe$98f40770$0300a8c0@pc52> 
Message-ID: <2893.1218795001@critter.freebsd.dk>

In message <04d601c8febe$98f40770$0300a8c0 at pc52>, "Tom Van Baak" writes:
>> I also maintain that the "right" forms of the zoneinfo files already
>> are proof of concept.  They demonstrate that computers can have an
>> uniform underlying time scale while still providing a civil time scale
>> that has leap seconds.  The technical issues are basically solved.
>
>I know I've been quiet on your zoneinfo thread last year, but
>my question is what percent of equipment or devices affected
>by leap seconds use zoneinfo files? 10%? 1%?

Since we don't even know how many systems are affected, because nobody
tries to count glitches, finding that number is pretty elusive.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at 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.

From jsm at polyomino.org.uk  Fri Aug 15 08:10:15 2008
From: jsm at polyomino.org.uk (Joseph S. Myers)
Date: Fri, 15 Aug 2008 12:10:15 +0000 (UTC)
Subject: [LEAPSECS] choose the form
In-Reply-To: <20080815060952.GA23990@ucolick.org>
References: <20080815060952.GA23990@ucolick.org>
Message-ID: <Pine.LNX.4.64.0808151158380.1254@digraph.polyomino.org.uk>

On Thu, 14 Aug 2008, Steve Allen wrote:

> I also maintain that the "right" forms of the zoneinfo files already
> are proof of concept.  They demonstrate that computers can have an
> uniform underlying time scale while still providing a civil time scale
> that has leap seconds.  The technical issues are basically solved.

As a proof of concept they tell us what *not* to do in any proper solution 
for transmitting both civil time and uniform time.  In particular, do 
*not* change the exported interpretation of past timestamps.  (For 
example, past POSIX timestamps stored in filesystems should remain 
interpreted as UTC according to the POSIX formula.)  As I said in my 
message of 11 February:

  The question does not relate to the *correctness* of those timestamps, 
  it's a case where consistency is much more important than correctness.

  If a system has a timestamp and exports an interpretation of that 
  timestamp, and another system stores that interpretation or a second 
  timestamp derived from it, and then one system changes its interpretation, 
  the timestamps are no longer in sync between the two systems.  This can 
  cause issues such as unnecessary remirroring of very large amounts of data 
  whose timestamps no longer match at the two ends.  See
  http://lists.debian.org/debian-user/1997/03/msg00075.html
  for an example of this from when Debian made the mistake of trying to go 
  it alone with a different interpretation of timestamps.  There are further 
  discussions in the debian-devel archives from March-May 1997 of things 
  that broke.

  If you do not change the timestamp interpretation formula in POSIX, but 
  stop inserting leap seconds at some point (or define that timestamps after 
  that point are in TI not UTC), you can avoid the issue of reinterpreting 
  existing timestamps.

So maybe something like "right", but only with a smooth transition from 
UTC to uniform time at some point well in the future.

-- 
Joseph S. Myers
jsm at polyomino.org.uk

From seaman at noao.edu  Fri Aug 15 12:38:08 2008
From: seaman at noao.edu (Rob Seaman)
Date: Fri, 15 Aug 2008 09:38:08 -0700
Subject: [LEAPSECS] choose the form
In-Reply-To: <2173.1218790615@critter.freebsd.dk>
References: <2173.1218790615@critter.freebsd.dk>
Message-ID: <63DDFAF2-D846-4837-97C4-7A743A3B5601@noao.edu>

Poul-Henning Kamp wrote:

> Thanks, a much needed conversion :-)

Steve is just saying what we've always said - there are two kinds of  
timescales and civil time needs to honor both.  It isn't the  
astronomers (also science software system engineers) who have failed  
to be open minded about how this is achieved.

> fiddling with it in hour-size bits every few hundred years is much  
> more likely to work than one second bits ever other year, so that  
> would be my preference.

Preference has nothing to do with it.  What are the requirements for  
the system?  To discover the requirements, consider the use cases.  To  
uncover the pertinent use cases, state the problem clearly.  Not one  
of these things has happened.  Hence candidate solutions to this  
unstated problem remain an issue of preference, not system engineering.

Rob Seaman
NOAO


From ashley at semantic.org  Sat Aug 16 03:02:57 2008
From: ashley at semantic.org (Ashley Yakeley)
Date: Sat, 16 Aug 2008 00:02:57 -0700
Subject: [LEAPSECS] choose the form
In-Reply-To: <20080815060952.GA23990@ucolick.org>
References: <20080815060952.GA23990@ucolick.org>
Message-ID: <1218870177.7710.3.camel@glastonbury>

On Thu, 2008-08-14 at 23:09 -0700, Steve Allen wrote:

> The technical issues are basically solved.

There aren't yet leap-second tables in tz, are there? This would be a
good step, or else an additional tz-like system for leap-second tables
and maybe other observed/decided information.

-- 
Ashley Yakeley


From cowan at ccil.org  Sat Aug 16 03:06:23 2008
From: cowan at ccil.org (John Cowan)
Date: Sat, 16 Aug 2008 03:06:23 -0400
Subject: [LEAPSECS] choose the form
In-Reply-To: <1218870177.7710.3.camel@glastonbury>
References: <20080815060952.GA23990@ucolick.org>
	<1218870177.7710.3.camel@glastonbury>
Message-ID: <20080816070623.GG9515@mercury.ccil.org>

Ashley Yakeley scripsit:

> There aren't yet leap-second tables in tz, are there? This would be a
> good step, or else an additional tz-like system for leap-second tables
> and maybe other observed/decided information.

Yes, there are.  Set your TZ variable to right/<whatever>.

-- 
Said Agatha Christie / To E. Philips Oppenheim  John Cowan
"Who is this Hemingway? / Who is this Proust?   cowan at ccil.org
Who is this Vladimir / Whatchamacallum,         http://www.ccil.org/~cowan
This neopostrealist / Rabble?" she groused.
        --George Starbuck, Pith and Vinegar

From sla at ucolick.org  Thu Aug 28 01:12:36 2008
From: sla at ucolick.org (Steve Allen)
Date: Wed, 27 Aug 2008 22:12:36 -0700
Subject: [LEAPSECS] but what does Daniel Gambis say?
Message-ID: <20080828051236.GA9422@ucolick.org>

Serbian Astronomical Journal, including mention of negative leap seconds

http://arXiv.org/pdf/0808.3612

--
Steve Allen                 <sla at ucolick.org>                WGS-84 (GPS)
UCO/Lick Observatory        Natural Sciences II, Room 165    Lat  +36.99855
University of California    Voice: +1 831 459 3046           Lng -122.06015
Santa Cruz, CA 95064        http://www.ucolick.org/~sla/     Hgt +250 m

From magnus at rubidium.dyndns.org  Thu Aug 28 01:50:05 2008
From: magnus at rubidium.dyndns.org (Magnus Danielson)
Date: Thu, 28 Aug 2008 07:50:05 +0200
Subject: [LEAPSECS] but what does Daniel Gambis say?
In-Reply-To: <20080828051236.GA9422@ucolick.org>
References: <20080828051236.GA9422@ucolick.org>
Message-ID: <48B63C8D.7020506@rubidium.dyndns.org>

Steve Allen wrote:
> Serbian Astronomical Journal, including mention of negative leap seconds
> 
> http://arXiv.org/pdf/0808.3612

Notice the detail that of the only projected leap second in the period 
2006-2024, it was predicted to 2008. This matches the achieved pattern 
so far.

It would be interesting to see how a prediction like this would turn out.

Cheers,
Magnus

From fmeyer at obs-besancon.fr  Thu Aug 28 07:06:13 2008
From: fmeyer at obs-besancon.fr (=?ISO-8859-15?Q?Fran=E7ois_Meyer?=)
Date: Thu, 28 Aug 2008 13:06:13 +0200 (CEST)
Subject: [LEAPSECS] but what does Daniel Gambis say?B
In-Reply-To: <20080828051236.GA9422@ucolick.org>
References: <20080828051236.GA9422@ucolick.org>
Message-ID: <alpine.DEB.1.00.0808281150170.7329@jupiter.obs-besancon.fr>

On Wed, 27 Aug 2008, Steve Allen wrote:

> Serbian Astronomical Journal, including mention of negative leap seconds
>
> http://arXiv.org/pdf/0808.3612

As the authors write, this is a pure mathematical
prediction ; of course the agreement between the
model and the data is excellent for the period
covered by data, but with 17 harmonic terms it could
hardly be different. Now, though it is hard in the
paper to figure it accurately, it is clear that the
extrapolated curve rapidly and significantly
diverges from the data as soon as it goes backwards
in time before the dataset used for the fit (<1962) : i
there is no stated reason that this could be different 
for the period after the data (after 2007).

So the paper conclusion would more qualify as a
very speculative bet than as an argumented 
prediction.

-- 
Fran?ois Meyer

From phk at phk.freebsd.dk  Thu Aug 28 07:11:14 2008
From: phk at phk.freebsd.dk (Poul-Henning Kamp)
Date: Thu, 28 Aug 2008 11:11:14 +0000
Subject: [LEAPSECS] but what does Daniel Gambis say?B
In-Reply-To: Your message of "Thu, 28 Aug 2008 13:06:13 +0200."
	<alpine.DEB.1.00.0808281150170.7329@jupiter.obs-besancon.fr> 
Message-ID: <40024.1219921874@critter.freebsd.dk>

In message <alpine.DEB.1.00.0808281150170.7329 at jupiter.obs-besancon.fr>, =?ISO-
8859-15?Q?Fran=E7ois_Meyer?= writes:

>Now, though it is hard in the
>paper to figure it accurately, 

Look at figure 4, the appropriate caption would have been "Our model
is bunk".

Apart from forcing some poor undergraduates though a lot of math,
this paper seems to have very little reason to exist.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at 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.

From tvb at LeapSecond.com  Thu Aug 28 07:43:21 2008
From: tvb at LeapSecond.com (Tom Van Baak)
Date: Thu, 28 Aug 2008 04:43:21 -0700
Subject: [LEAPSECS] but what does Daniel Gambis say?B
References: <40024.1219921874@critter.freebsd.dk>
Message-ID: <003601c90903$48c4d1e0$0700a8c0@pc52>

To be more convincing I'd like to see what happens to their
model if they used 5, 10, or 15 terms instead of 17. Plotting
the quality of fit against the number of terms used would be
revealing.

Also I'd like to see what happens to their projections if they
used a shorter or longer range of dates. It must be possible
to fit any data set given enough terms so the question is how
sensitive the parameters of the fit are.

Future data isn't available so they'd have to work with smaller
segments from the past to pull this off. For example, they used
about 44 years of data (1962 to 2006); how well would their
model work if they used only 10 or 20 or 30 years instead of
44? Plotting quality of fit against the number of years used
would be interesting. Or fairer yet, against the ratio of years
of data in the past vs. years of prediction in the future.

/tvb


From clive at demon.net  Thu Aug 28 07:30:47 2008
From: clive at demon.net (Clive D.W. Feather)
Date: Thu, 28 Aug 2008 12:30:47 +0100
Subject: [LEAPSECS] but what does Daniel Gambis say?B
In-Reply-To: <alpine.DEB.1.00.0808281150170.7329@jupiter.obs-besancon.fr>
References: <20080828051236.GA9422@ucolick.org>
	<alpine.DEB.1.00.0808281150170.7329@jupiter.obs-besancon.fr>
Message-ID: <20080828113047.GO319@finch-staff-1.thus.net>

Fran?ois Meyer said:
> http://arXiv.org/pdf/0808.3612
> 
> As the authors write, this is a pure mathematical
> prediction ; of course the agreement between the
> model and the data is excellent for the period
> covered by data, but with 17 harmonic terms it could
> hardly be different.

How many data points are they using to get the fit? With 17 harmonic terms
they have 53 degrees of freedom in their equation.

This feels very much like a case of adding too many terms to your
equation so that it's bound to fit the data. I've had some nasty
experiences because someone decided to fit a fifth-order polynomial to some
data without looking at the results (electrical efficiency was swinging
between -300% and +400%).

-- 
Clive D.W. Feather  | Work:  <clive at demon.net>   | Tel:    +44 20 8495 6138
Internet Expert     | Home:  <clive at davros.org>  | Fax:    +44 870 051 9937
Demon Internet      | WWW: http://www.davros.org | Mobile: +44 7973 377646
THUS plc            |                            |

From seaman at noao.edu  Thu Aug 28 16:21:10 2008
From: seaman at noao.edu (Rob Seaman)
Date: Thu, 28 Aug 2008 13:21:10 -0700
Subject: [LEAPSECS] but what does Daniel Gambis say?
In-Reply-To: <20080828113047.GO319@finch-staff-1.thus.net>
References: <20080828051236.GA9422@ucolick.org>
	<alpine.DEB.1.00.0808281150170.7329@jupiter.obs-besancon.fr>
	<20080828113047.GO319@finch-staff-1.thus.net>
Message-ID: <48BE5E1D-7E1C-4F15-8D8D-47DC309B2A34@noao.edu>

> Serbian Astronomical Journal, including mention of negative leap  
> seconds
>
> http://arXiv.org/pdf/0808.3612

Timekeeping is one of those narrow disciplines where peer review  
implies small number statistics.  So this paper joins others of  
suspect validity such as:

	Deines & Williams (2007, AJ, 134, 64) "Time Dialation and the Length  
of the Second: Why Timescales Diverge"

That said, it doesn't appear to be devoid of content - simply rather  
"lazy".  I'm particularly taken with statements like "all the physical  
factors unambiguously recognized till nowadays cannot cause such a  
phenomenon".  Most journal editors and referees would have a hard time  
accepting a paper with such a conclusion.  Rather, the authors'  
curiosity should have driven them to further investigation of one sort  
or another.  Either tie the results to the current physical basis of  
the field, or acquire data sufficient to call current physical  
understanding into question.

One is left wondering how this paper came to be written.  PHK's  
supposition about undergrads sounds about right.  Some of the choice  
of wording seems to imply that the authors are directly or indirectly  
aware of discussions from this mailing list.  The title suggests an  
interest in abetting the ITU-R's agenda.  The shallow analysis (and  
obscure journal) should make this all rather moot.  I am pleased,  
however, to see some attempt at predicting Earth orientation, but this  
surely cannot represent the state of the attainable art :-)

I can't add much to the previous comments that a fit with so many  
degrees of freedom is immediately suspect for purposes of  
extrapolation.  It is unremarkable that they were able to track every  
tiny wiggle in the data.  FIgure 4 is quite humorous.

Modeling delta-T, a cumulative effect, doesn't seem very expressive in  
general - all the empirical insight was smoothed out of the data  
before they started.  Rather, it would be better to focus on LOD or  
other more direct indicators of Earth orientation or rotation rate.   
Integrate the models to recover UT1, etc., as a separate step.  A fit  
without residuals is like signal with no noise - begging for skepticism.

One could seek physical explanations for the more (statistically)  
significant harmonic terms.  For example, the 22 year period is twice  
the solar cycle and there could be coupling with the magnetic field  
reversals.  A few of the terms in this (apparently) unconstrained fit  
pop out near annual, fractional annual, and multi-annual periods.   
Probably not a coincidence.  The solar system is rife with harmonic  
resonances of various sorts.  Surely the Earth rotation community must  
have considered a number of such phenomena previously given that lunar  
tidal locking is recognized as the principal driver for Earth's  
rotational slowing?

On the other hand, understanding the short period seasonal effects on  
LOD (all those beautiful wiggles from the plots on Steve Allen's web  
site, for instance) benefits simply from coverage over many cycles.   
It has taken a century or two to figure out the solar cycle due to its  
long period.  Long term LOD phenomena will take similarly long to sort  
out.  A comparative study with LOD data from the other planets and  
satellites could be revealing.  (One suspects the available data  
aren't sufficiently precise for the other planets.  Sounds like a good  
argument to send astronomers to Mars :-)

The 222 year period is most likely a meaningless artifact of the  
data.  The long term trend of tidal slowing is going to pop out as the  
most significant term.  The corresponding period is truncated by the  
truncated horizon of the available data.

Would prefer to see a similar fit with fewer terms and more  
constraints.  How about forcing terms with periods of a lunar month, a  
solar year, etc?  In addition to the free harmonic terms, how about an  
explicit linear trend forcing the long term tidal slowing?  How about  
forcing the millennial term from http://www.ucolick.org/~sla/leapsecs/ancient.pdf? 
   How about estimating the magnitude of the coupling between the  
terrestrial magnetic field and the solar magnetic field?  Could this  
possibly transfer enough angular momentum to affect the Earth's  
rotation?

This particular paper is only worth attention due to the dearth of  
similar papers.

Rob Seaman
NOAO
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://six.pairlist.net/pipermail/leapsecs/attachments/20080828/2751132f/attachment.htm>

From phk at phk.freebsd.dk  Thu Aug 28 16:35:13 2008
From: phk at phk.freebsd.dk (Poul-Henning Kamp)
Date: Thu, 28 Aug 2008 20:35:13 +0000
Subject: [LEAPSECS] but what does Daniel Gambis say?
In-Reply-To: Your message of "Thu, 28 Aug 2008 13:21:10 MST."
	<48BE5E1D-7E1C-4F15-8D8D-47DC309B2A34@noao.edu> 
Message-ID: <42686.1219955713@critter.freebsd.dk>

In message <48BE5E1D-7E1C-4F15-8D8D-47DC309B2A34 at noao.edu>, Rob Seaman writes:

>The 222 year period is most likely a meaningless artifact of the  
>data.

Likely lack of DC removal and/or deficient windowing before running
the FFT I suspect they used.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at 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.

From seaman at noao.edu  Thu Aug 28 16:49:12 2008
From: seaman at noao.edu (Rob Seaman)
Date: Thu, 28 Aug 2008 13:49:12 -0700
Subject: [LEAPSECS] but what does Daniel Gambis say?
In-Reply-To: <42686.1219955713@critter.freebsd.dk>
References: <42686.1219955713@critter.freebsd.dk>
Message-ID: <E5E0F161-EF7B-4B6C-B015-64543F953EF1@noao.edu>

Right.  That's a better way of saying it.  The paper doesn't describe  
the numerical techniques well enough to critique them.  There is some  
sort of linear component to the fit, but it doesn't appear to  
correspond to the tidal slowing term.  Perhaps clues are buried in the  
references.
--

On Aug 28, 2008, at 1:35 PM, Poul-Henning Kamp wrote:

> In message <48BE5E1D-7E1C-4F15-8D8D-47DC309B2A34 at noao.edu>, Rob  
> Seaman writes:
>
>> The 222 year period is most likely a meaningless artifact of the
>> data.
>
> Likely lack of DC removal and/or deficient windowing before running
> the FFT I suspect they used.
>
> -- 
> Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> phk at 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.
> _______________________________________________
> LEAPSECS mailing list
> LEAPSECS at leapsecond.com
> http://six.pairlist.net/mailman/listinfo/leapsecs