[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] TAI epoch

On Sun, 2014-03-23 at 19:25 -0600, Alan Watson wrote:
> > R7RS did not invent the POSIX 1970 + 10 second epoch.
> > The question is why did older libraries use this (and does
> > anybody know)?
> I don’t know for sure, but I’d guess that because prior to 1972 atomic
> seconds and astronomical seconds were different, so the difference
> between atomic time and astronomical time was not an integral number of
> seconds. The earliest epoch for which the common atomic time scale
> coincides with the common astronomical time scale to within an integral
> number of seconds is 1972, at which point the difference was 10
> seconds.

Okay, terminology.  Is the "astronomical second" you're referring 
to the exactly-equal-to-SI-second exactly 31557600 of which
subdivide the time required for light to travel one standard 
light year, or a running count of those exactly-equal-to-SI seconds?  

Or is it the not-quite-equal-to-SI and also not-quite-constant and 
also not-quite-predictable time unit exactly 86400 of which 
subdivide the time required for Earth to make exactly one complete
rotation relative to the sun, or a running count of those inexactly-
known seconds (which most "naive" time handling applications assume, 
causing bugs when run against system clocks that occasionally put 
61 seconds into a minute)?

Or is it the running count of the second type, plus the running count 
of "leap seconds" that are supposed to adjust the running count of 
the second type so that it advances congruently with the first?

Scheme-reports mailing list