[LEAPSECS] Terminology question

[M. Warner Losh]

In message: <20100309173840.GB830 at ucolick.org>
            Steve Allen <sla at ucolick.org> writes:

: On Tue 2010-03-09T17:07:01 +0000, Tony Finch hath writ:

: > Is there a generic term for timescales like POSIX time_t and NTP that

: > count seconds (or some other interval) since an epoch without taking

: > leap seconds into account?

: 

: mean solar time

: 

: thus, by action of the International Meridian Conference and the IAU

: and the CGPM,

: 

: Universal Time


Are you sure that mean solar time is the right way to describe these
time scales?  You can measure mean solar time with a count of actual
seconds since some epoch with whatever time adjustments needed to keep
them in sync with mean solar time.  You'd use the adjustments when
converting from the "long count" of seconds to a broken down time
that's defined, in say, the UTC documents (ITU TF.460-4).  This way
you can convert a number to the corresponding UTC time.  With the
extra tables, you can even get the 23:59:60 bit right, which is
impossible with time_t since (a) that second isn't well defined (is it
a repeat of the last second of the day of the leap second, or a repeat
of the first second of the following day) and (b) ambiguous.

NTP and time_t ignore the leap seconds so that it is possible to
convert the long count of seconds into a broken down time without the
use of tables.  They do this by pretending leap seconds don't exist
(NTP kinda knows about them, but the long count of seconds assumes
they don't exist).  I'm not familiar with a term that describes this
kind of arrangement.

I guess it depends on what properties of time_t and NTP counts you are
looking to highlight.

Warner