Subject: Re: Lisp Users Group Meeting
From: Erik Naggum <>
Date: 1999/10/18
Newsgroups: comp.lang.lisp
Message-ID: <>

* Gareth McCaughan <>
| Presumably it's complicated by leap seconds.  But then, so is any other
| representation. It sounds pretty good to me.

  it is important to understand that leap seconds take two forms:

A separately identified seconds (23:59:59, 23:59:60)
B seconds that last twice as long (23:59:59, 23:59:59)

  the former applies to what I call "Scientific Time" in my paper, the
  latter to what I call "Political Time".  the scientific community is
  obviously concerned with well-behaved time, while the political
  community is concerned with well-behaved people through well-behaved,
  yet manually updated, clocks.  Scientific Time is fundamentally relative,
  while Political Time is fundamentally absolute, although at first glance
  it appears the opposite must be true.  Scientific Time maintains the
  concept of time differential as its primary concern, while Political Time
  is concerned with points in time, _only_, and is oblivious to the effects
  of points in time that have no well-defined differential between them.

  when dealing with Political Time, it is therefore meaningless to try to
  capture leap seconds as no actual clock will show the time that the
  scientific community will want to label the leap second.  mapping between
  Scientific Time and Political Time in this regard is also trivial, and
  protocols such as the Internet Network Time Protocol are very good at
  this.  there is, in consequence, no leap second to be concerned with.

  put another way, I called my time concept "local time" because it applies
  to what time is perceived to be locally.  "universal time" is just that,
  and is furthermore _unsuited_ to represent local time as humans want to
  experience it.  universal time may or may not be suited to represent
  "real time" or other concept of time differentials -- I have not looked
  into that issue as deeply as it would require to take a firm position.

  the paper is available at <URL:>, and
  is about 60K large.  the link to the source code and implementation is
  not yet operational.