use-revolution digest, Vol 1 #332 - 13 msgs

[nedlud at postoffice.pacbell.net]

January 22, 1998
From: Jonathan <nedlud at pacbell.net>
To: <use-revolution at lists.runrev.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"

Rob, 

I'm afraid you are mistaken. Seconds are extremely limited. Their limits 
are directly related to the limits of the computer. At the moment this 
means that time starts in the 1970's. So, for example, if you want to do 
common calculations such as generating the age of a person from their 
birthday you are screwed. The same is true for any calculation outside of 
the operating systems time window. Furthermore, I suspect that this is 
different on different platforms, so the problem will not be consistent. 
Julian dates are the standard for long time periods. Furthermore, Julian 
arithmetic is far simpler than the calculations required for seconds (no 
multiplication needed in most cases). And there are convenient methods 
for determining the day of the week -- something that is useful for 
calendars. Julian system is also a more compact representation -- seconds 
will eventually overflow when dealing with large time differentials. 

The closest functionality would be the date items function, but that is 
also tied to the computer clock and is therefore similarly limited. 

Frankly, I strongly believe that Julian dates should be the internal 
format since that would provide us with a standard time measure that is 
truly standard across all systems.

Also, I believe that Richard Gaskin posted a set of julian functions to 
the metacard list 3 or 4 months ago.

J/



Rob Cozens <rcozens at pon.net> wrote on 4/10/2002 11:49 AM

>The real bottom line is there are internal date representations (eg: 
>seconds; the long date) that naturally sort in chronological order, 
>and using seconds one can add/subtract days (1 day = 86400 seconds); 
>so there is little practical need for Julian date representation.
>--