[LEAPSECS] Testing computer leap-second handling

[Michael Spacefalcon]

Poul-Henning Kamp <phk at phk.freebsd.dk> wrote:


> Yes +/- 4 hours or so.


Only for you and those of your ilk.  For me the tolerance is much,
much tighter:

* The long-term measure of days must agree with the true evolution of
  mean solar time (longitude-independent) to within 1 s.

* Using the mean solar time at a standardized time zone longitude
  instead of that at the exact longitude of your house is OK as long
  as the difference does not exceed 30 min.  This rule implies that
  most communities can make do with integral hour offsets from GMT (if
  each zone is centered on an exact hour longitude, no point in the
  zone is more than 30 min off), but some communities may need to use
  half-hour zones.  If a community happens to lie right in between
  one-hour longitude marks, with integral hour GMT offsets they can
  either have the boundary cut through the community (very inconvenient)
  or pick one side or the other.  But with the latter approach, those
  citizens who happen to be on the wrong side will have their fundamental
  human rights violated by being subjected to a delta between true MST
  and civil time than exceeds 30 min, hence that approach is impermissible.
  Such communities need half-hour time zones.

* DST is an abomination and impermissible.  A location's longitude
  remains unchanged all year round, and because a civil time can remain
  compliant with the basic human right only as long as it remains in
  a tight tolerance of the true MST, DST cannot be implemented without
  trampling over human rights.

SF