I've been reading the list archives. Parts of the discussion are rather
repetetive. I think the search space could be narrowed quite a bit if the
list produced a canonical statement of consensus, listing facts on which
there is no dispute. This would serve much the same purpose as a FAQ,
as well as possibly a base for a final report on the leap seconds issue.
Let us grok in fullness before coming to a conclusion.
So as a start here are some statements that I think are not controversial
on this list. If any are disputed, please speak up.
nature of time scales
---------------------
UT1 et al are not really measures of time, but of angle (of Terran
rotation).
Readings of UT1 et al are most naturally represented as a real count of
rotations since some epoch (i.e., as some form of Julian Date).
Because TT, TAI, et al are measures of time unrelated to planetary
rotation, it is misleading to apply to them the day-based notations
(such as the sexegesimal time-of-day notation) that are customarily used
with UT1 et al.
Readings of linear time scales (TT, TAI, et al) are most naturally
represented as a real count of SI seconds since some epoch.
Post-1972 UTC, counting TAI seconds while maintaining a "day" cycle that
closely matches the phase of UT1, is directly analogous to calendars
that count days while maintaining a "year" cycle that closely matches
the phase of the tropical year.
Readings of UTC cannot be directly represented by a single linear count.
As a calendar, UTC is presently of the observational variety.
civil time
----------
Up to the present, most human activity has been in the long term
synchronised with local solar time.
We are not in a position to determine whether, to what extent, and for
how long, the synchronisation between activity periods of Terran humans
and the rotation of the planet will be maintained.
Up to the present, local civil time has approximately maintained a
conventional correspondence with the timing of human activity.
By the use of standard time zones, the correspondence between local
civil time and local mean solar time has historically varied within a
range of about five hours, excluding arctic regions.
It is presently commonplace for the correspondence between local civil
time and local mean solar time to vary periodically with an amplitude
of one hour.
It is convenient for the civil time in different localities to have a
simple relationship.
Where civil time involves a unit very close in duration to the SI second,
it is convenient for that to actually be the SI second on the geoid.
time handling in software
-------------------------
Unix time, as standardised by POSIX and as commonly implemented, is a
faulty encoding of UTC. The fault is that Unix time readings repeat,
and so are ambiguous, near positive leap seconds.
The Unix time interfaces are capable of being used to correctly
encode UT1, TAI, UTC-SLS, or other time scales that do not have
internally-visible leap seconds.
Some applications assume that Unix time is monotonically nondecreasing,
or that timestamps are unambiguous, and so are poorly served by the
encoding of UTC in Unix time.
Some applications assume that Unix time is a linear scale suitable for
interval calculations, and so are poorly served by the encoding of any
form of UT in Unix time.
Some applications assume that Unix time encodes UTC, including
discontinuous behaviour around leap seconds, and so would be poorly
served by the encoding of anything other than UTC in Unix time.
New applications need a more sophisticated understanding of time than
is currently standard practice.
Various applications require means to handle civil time, TAI, and UT1,
among other time scales.
Applications need to process times that are contemporary, historical,
and in the foreseeable future, on all time scales they are interested in.
When dealing with observational calendars and other unpredictable
aspects of time scales, applications need a way to be sure that they
have sufficiently fresh information about calendar structure.
time dissemination
------------------
Human society needs to develop better means to disseminate a multiplicity
of time scales.
-zefram
Received on Thu Jun 01 2006 - 04:07:26 PDT
This archive was generated by hypermail 2.3.0
: Sat Sep 04 2010 - 09:44:55 PDT