[LEAPSECS] Terminology question

[M. Warner Losh]

In message: <4B984285.3000004 at yahoo.com>
            Michael Deckers <michael.deckers at yahoo.com> writes:

: >  In the lingo of the atomic horologists I would say

: >  "the relationship between UTC(TAI) and TAI is simple."

: >

: >  Here UTC(TAI) means "the version of UTC constructed

: >  in arrears by using the contents of Circular T".

: 

:    But IERS Bulletin C would suffice for the

:    relationship between UTC and TAI, and that is

:    available a bit earlier.


Yes.  If you have a decent UTC(x), that's traceable to TAI, then you
have a fairly good idea what your delta will be.  UTC(NIST), for
example, is on the order of tens of nanoseconds.  For most
applications, this small variation isn't worth bothering over.  But
for some it may be critical.  UTC(x) is always going to be a real-time
approximation of UTC, which is always N seconds behind TAI.

What this has to do with time scale representation that allows for
naive math to produce displayable results without the need for a
leapsecond table, however, is beyond me. :)  Here, btw, naive math
means that you can get a count of the days since 1970 by taking the
time_t value / 86400.  You can get the second within the day with
value % 86400, etc all the way down to fully broken down time.  The
algorithm is deterministic, and needs no tables to do the conversion.

The talk of UTC here really isn't the same thing as time_t.  Time_t is
an encoding of the broken down time of UTC that is reversible and
deterministic, but one that blithely ignores that leap seconds exist.
So you can take a UTC time (2010-03-10 10:10:10) and convert it to a
time_t and back again for all values except :60.  It is important to
note that the UTC time scale is usually measured as day (either JD,
MJD or Gregorian break down) and time within the day (either as a
fraction of the day, or a count of second of day, or as a hh:mm:ss
time).  The MJD variation of UTC is completely convertible to the
official form of UTC, and makes delta t measured in days easy to
compute.

Of course, after saying all that, I'm left with no better term that's
in wide spread use, so I have to resort to labels like "pseudo-linear"
or "naive-math friendly".  But maybe we need to borrow from my college
days in physics class and call it an "idealized time count" or
something like that to show that it is a polite fiction that makes the
math easier and mostly right most of the time...

Warner