[ntp:questions] Re: no more leap seconds?
Richard B. Gilbert
rgilbert88 at comcast.net
Thu Aug 11 03:50:25 UTC 2005
Rich Schmidt wrote:
>Then in Oct. 2004 the US Working Group 7A submitted a proposal to
>eliminate leap seconds in Dec. 2007, and to introduce leap hours, the
>first of which would not be needed for about 600 years (based on
>current analysis of historical observations of the Earth's rotation
>rate). As with leap seconds, leap hours would not be regularly
>scheduled but only as needed, and 600 years would almost give enough
>time for programmers of the world to work leap seconds into their code.
Let's be realistic about this. If the first leap hour does not occur
for 600 years, programmers will not begin to consider the problem until
about 2603 AD. Can you say "Y2K all over again"? I saw an article
or letter to the editor defining the Y2K problem in, I think, Datamation
in the 1970's. This was at least twenty years before the fact. When
did anybody actually do anything about it? Most shops did not even
commence efforts to test and fix their software until late 1998 or early
1999. Would *you* invest a great deal of money in an effort that would
show no return for 600 years?
By the time 2603 rolls around, there will be will 10^20 lines of code
that will need to be modified. The source for at least half of the
affected code will be incomplete or missing altogether. The problem
will be worked around by agreeing to add an extra leap day 12,000 years
in the future.
The system we have may, indeed, be flawed but it has worked "well enough".
More information about the questions