[LEAPSECS] Civil timekeeping requirements versus UTC specifications

[Rob Seaman]

Steve is discussing engineering requirements - that is, a description of the problem and its context in the real world.  Tony and Warner are discussing specifications for particular ("specific") candidate solutions.  See http://www.cacr.caltech.edu/futureofutc/preprints/02_AAS_11-661_Seaman.pdf or any number of system engineering references.  System engineering best practices then supply tools for evaluating a number of proffered candidate solutions one against the other and against cost, schedule, performance and risks.

However the "Draft Revision to ITU-R Recommendation TF.460-6" discusses none of this.  My interpretation of Tony's message here is that he is attempting to dislodge Steve's (rock solid) assertion of a requirement.  Warner on the other hand is exploring the solution space as an iterative process for better understanding the problem space.  But the ITU is seeking to sweep the whole thing under a rug that is two sizes too small.

Rob Seaman
National Optical Astronomy Observatory
--
On Jan 4, 2012, at 12:58 PM, Warner Losh wrote:


> On Jan 4, 2012, at 12:45 PM, Tony Finch wrote:

> 

>> Steve Allen <sla at ucolick.org> wrote:

>>> 

>>> Happy New Year!

>>> 

>>> How did I tell that it became the new year?

>>> By the civil calendar date changing from Dec 31 to Jan 1.

>>> How did I tell when the civil calendar date changed?

>>> By looking at my clock set using UTC, which is, at present, connected

>>> to the rotation of the earth, and thus to days.

>>> The draft revision of ITU-R TF.460 removes the connection between UTC

>>> and the rotation of the earth,

>> 

>> Correct up to this point.

>> 

>>> thus disconnecting the civil calendar date from the sun.

>> 

>> Incorrect, because the civil calendar is tied to local solar time using

>> the time zone system. Local civil time is not going to move out of sync

>> with the sun, and therefore the calendar will also remain in sync.

> 

> Do people have a notion how we'll recon time when the accumulated delta becomes large (like on the order of 100k seconds)?  UTC+27?

> 

> Warner