[LEAPSECS] Standards of time zones -Brooks Harris

Rob Seaman seaman at noao.edu
Fri Jan 10 10:45:31 EST 2014


On Jan 10, 2014, at 2:39 AM, Poul-Henning Kamp <phk at phk.freebsd.dk> wrote:


> You mean these people ?

> http://www.antipope.org/charlie/blog-static/2013/12/metaphor-for-the-day.html


The "metaphor for the day" tag had my hopes up that this would be a metaphor for the concept of "day"...no such luck. Not sure one should put much faith in a religious metaphor from a site called "antipope.org" :-) That said, POSIX is more like the Illuminati...or perhaps the Spanish Inquisition as in one of the comments. Comparing it to the clarity and poetry of the Nicene Creed is absurd (and rather offensive).

It commits the sin of despair to suggest that computing is too splintered to ever again have a prayer of putting things right. Rather, there have been ecumenical discussions throughout the history of Unix - for example my own modest experience of porting a large astronomy package to a "dual-universe" SysV/BSD hybrid in the mid-1980s.

Future community infrastructure - future computing communion, if you will - cannot be built on faulty physical models of the universe.


On Jan 10, 2014, at 2:49 AM, Poul-Henning Kamp <phk at phk.freebsd.dk> wrote:


> In message <20140110064412.GB20400 at ucolick.org>, Steve Allen writes:

>

>> I have asserted that POSIX does not want to know astronomy, does not

>> want to track geophysics, [...]

>

> You are right about that.

>

>> that POSIX really wants to count atomic days

>> rather than mean solar days,

>

> And wrong about that.

>

> POSIX want to track the timescale underlying civilian and legal

> timekeeping, the timescale known as UTC, aka GMT, aka Worldtime,

> aka Weltzeit etc, because computers are used to implement the

> civil and legal society.


Like I said, "ecumenical". Civil timekeeping depends on both. "Day" is defined for us since we live on planet Earth (http://futureofutc.org/preprints/files/28_AAS_13-515_Seaman.pdf). The SI unit of duration - really a unit of frequency - happens to have been given a confusing name and 1820 calibration, but is a separate concept. Humans rely on both the day and the second on a daily basis. "Second" happens to have an overloaded meaning; many concepts do.

That POSIX does a poor job of separating the two notions is not a coherent argument for seven billion people (and untold generations to come) to attempt to pretend that day means 11.57407 microhertz.


> Given sufficient resources, nothing prevents us from fixing this

> "the right way", pressuming we can ever agree what that is.

>

> Unfortunately "sufficient resources" are not available, not even close.


The implicit assertion here is that it will be cheaper - and inerrantly safe - to attempt to sweep the issue under the rug. Rather, any coherent decision-making process would quantify the costs and risks of all options. There is no rug big enough.

There are also more options than the ITU's naive proposal to terminate leap seconds and be damned to the consequences (http://en.wikipedia.org/wiki/Law_of_the_instrument).


> Therefore people, primarily the US DoD who has a lot of shitty

> software of that vintage, including appearantly the entire control

> system for at least one part of the nuclear triad, are pushing the

> cheapest and simplest solution they can find: "Drop leap seconds".


Rather, the DoD "position" is the result of lobbying of an obscure office (that no longer exists) by a small group of self-interested parties. Ceasing leap seconds carries risks that require evaluation. Coherent systems engineering (rising perhaps to the level of operations research) is needed to even comprehend the problem space, let alone entertain a proposed substitute "solution".


> The only people identified which really care abut DUT1 are people

> who point telescopes and dishes at extraterrstial objects.


Those who navigate spacecraft, submarines, aircraft, ships and ground transport and who deployed GPS in support of these also care. For instance, regarding the relaxation of the DUT1 0.9s limit:

"While eliminating this ‘sanity check’ may seem trivial, changes to the code, documentation and execution of thorough testing will require resources and in some cases, contract modifications. The costs and time needed for the initial investigation and subsequently to make the required changes to the operational software are unknown at this time, but they are expected to be significant. Furthermore, considering that this possible redefinition of UTC and the elimination of the leap second offer no benefits to NGA GPS operations and GPS users, pursuit of such a fundamental change appears to be an inefficient use of limited resources."

(http://www.cacr.caltech.edu/futureofutc/2011/preprints/32_AAS_11-675_Malys.pdf)

More broadly:

"The proposed discontinuation of leap seconds and redefinition of UTC will impact the operational software and automated transfer of Earth Orientation Prediction Parameters between NGA, the GPS OCS, and other DoD organizations. A significant amount of time, effort, and funding will be required for NGA and other organizations to identify and assess all operational software that references, tests for, or applies the UT1-UTC parameter in high-accuracy orbit determination processes. The costs and time needed for the required changes to the operational software and ICDs are unknown at this time, but they are expected to be significant.

"While the proposal to re-define UTC may offer benefits to other communities, a redefinition of UTC and the resulting elimination of leap seconds offer no benefit or improvement to NGA or GPS operations. Our recommendation, therefore, would be to maintain the current definition of UTC."

Rob



More information about the LEAPSECS mailing list