[time-nuts] Re: UTC - A Cautionary Tale
David Forbes
dforbes at dakotacom.net
Thu Jul 14 12:26:20 EDT 2005
Brooke,
True. However, the timescale in which Leap Hours are interesting is
also that in which 5 digit years are required, so those problems can
both be fixed by the COBOL programmers in the late 9990s. [humor]
GPS time is not a thing that astronomers want to use in the long run,
although they currently use GPS receivers to set their clocks to UTC
and derive sidereal time from that. [humorous but true]
The problem with using GPS time is that most GPS receivers convert it
to UTC before giving it to you!
At 9:03 AM -0700 7/14/05, Brooke Clarke wrote:
>Hi David:
>
>My concern is that if the time between leaps gets to be long then
>there will be another Y2K type problem. I.e. programmers will
>ignore the Leap Hour, figuring that they will be dead when it
>occurs, and when it does there will be many broken programs.
>
>The GPS time scale does not have leap seconds. Would it be suitable
>for those applications where leap seconds are a problem?
>
>Have Fun,
>
>Brooke Clarke, N6GCE
>--
>w/Java http://www.PRC68.com
>w/o Java http://www.pacificsites.com/~brooke/PRC68COM.shtml
>http://www.precisionclock.com
>
>David Forbes wrote:
>
>> At 11:13 PM -0700 7/13/05, Rob Seaman wrote:
>>
>>> Howdy,
>>>
>>>> This is a little missive from an astronomer on the delicate
>>>>subject of the divergence of UTC from UTx. It seems that those
>>>>bastards in the precision timing community want to abandon UTC's
>>>>leap seconds entirely because they are too much trouble, and he's
>>>>hopping mad.
>>>
>>>
>>> Note that my message was composed for astronomers, not you guys.
>>> Several of us in the astronomical software community have been
>>>following this issue since before Y2K:
>>>
>>> http://www.ucolick.org/~sla/leapsecs
>>>
>>> We are as "hopping mad" about the sneaky process as about the
>>>proposal. Note our two tiered objection: they not only propose
>>>to cease issuing leap seconds, they propose to continue calling
>>>the resulting time scale "Coordinated Universal Time". There are
>>>many flavors of UT - UTC should not be divorced from the others.
>>>Call a leap second-less civil time anything you want - simply
>>>don't call it "UTC".
>>
>> I agree that important processes should not be sneaky, but they
>>often are. Manhattan Project, anyone?
>>
>>>> [His most amusing argument against modifying UTC is that
>>>>astronomy software tends to use UTC not UT1 etc.]
>>>
>>>
>>> Amusing how?
>>
>> It's amusing in that UTC is civil time, not astronomical time,
>>which one would expect astronomers to use. I didn't say it's bad or
>>wrong, just that it's amusing. Jokes are amusing. I have a sense of
>>humor, which many people seem to lose when their favorite ideas are
>>attacked.
>>
>>> Also note that UT1 is only available after the fact. UTC is a
>>>deterministic (if segmented) timescale which provides not only an
>>>approximation (and prediction) of UT1, but also provides access to
>>>TAI two or three orders of magnitude more precisely yet. It may
>>>not be perfect, but then - this proposal isn't designed to provide
>>>something better. Imagine what might have been achieved if the
>>>precision timing community had spent the seven year leap second
>>>hiatus working to improve UTC rather than to sabotage it.
>>
>> UTC is NOT deterministic. It has leap seconds inserted randomly
>>with only 6 months advance notice. You can't plan a mission to
>>Saturn based on UTC.
>> There was a big discussion about this subject on the time-nuts
>>list a couple weeks ago precisely *because* UTC is not
>>deterministic. Computer programmers have to stand on their heads to
>>design systems to calculate future time using UTC.
>>
>>> I find it surreal that it is the precision timing community who
>>>are arguing that the public have no need for access to precision
>>>time.
>>
>> The time the public uses doesn't need to be locked to the Earth's
>>rotation to within a second over the short term. The thing to solve
>>is the long-term drift, which can be predicted far in advance, but
>>not to within a second a year.
>> I propose a better solution that will keep the civil timescale
>>locked to the Earth's rotation to within a minute and be
>>deterministic for hundreds of years in advance: Create leap minutes
>>and *define them in advance* for the next 500 years (or however far
>>in advance is practical) based on the second-order curve of the
>>known characteristics of the Earth's rotation. Then the programmers
>>will have an algorithm to guarantee that their clock code will work
>>until long after they're dead.
>>
>>> Rob Seaman
>>> NOAO
>>>
>>> _______________________________________________
>>> time-nuts mailing list
>>> time-nuts at febo.com
>>> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>
>
>
>
>_______________________________________________
>time-nuts mailing list
>time-nuts at febo.com
>https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
--
--David Forbes, Tucson, AZ
http://www.cathodecorner.com/
More information about the time-nuts
mailing list