Date: Tue, 14 Dec 2010 13:59:23 -0500
   From: John Cowan <cowan-x6Fv9fcoW1Y+uZM1RHzMvw@public.gmane.org>

   I have added a second return value to TimeCowan's current-posix-seconds
   which is 0 if this is definitely not a leap second, 1 if it is definitely
   a leap second, and #f if the implementation doesn't know.  Does that
   make you sufficiently happy?

This is absurd.  Programs dealing with timing shouldn't have calendars
thrust at them.  Give them clocks, not calendars.  Count the number of
SI seconds since the UTC epoch.

Calendars, requiring such context as time zones for local civil time
and solar observations for leap seconds in UTC and lunar observations
for the Islamic calendar and so on, are useful -- for applications
that process calendars.  Make the calendars aware of leap seconds, not
the clock.