[LEAPSECS] No leapseconds on trains

[Zefram]

Paul J. Ste. Marie wrote:

>This is abundantly clear


Clear, but in the POSIX standard it's immediately redefined to mean
something other than what it clearly says.  It defines "seconds since
the epoch" to be a numerical encoding of UTC timestamps, *not* a linear
count of seconds.  The MVB standard seems to be best read as incorporating
that definition.


>                          Absolute number of SI seconds since  

>1970-01-01Z00:00:00 doesn't change with or without leap-seconds.


Yes.  However, 1970-01-01 is in the rubber-seconds era of UTC.  So, if
taking it literally, are the "seconds" meant to be UTC seconds or TAI/SI
seconds?  It doesn't say which, and there are more of the latter than of
the former.  Either way, the number of seconds from 1970-01-01T00:00:00.0
UTC to (say) 2011-01-01T00:00:00.0 UTC is not an integer.  I don't think
*anyone* genuinely counts seconds from 1970-01-01T00:00:00 UTC.

-zefram