[James Edward Gray II <james / grayproductions.net>, 2004-08-27 16.07 CEST]
> On Aug 27, 2004, at 8:43 AM, Carlos wrote:
> 
> >Anyway, don't use Time: you will have problems when you add 86400 
> >seconds to
> >advance a day if it's near midnight and on the day when the summer time
> >changes.
> 
> You bring up a good point, but is this true...
> 
> Time keeps track of it's value in seconds since the Epoch right?  Then 
> just conveniently answers questions about what day that number of 
> second would fall on, well if I understand it correctly anyway.
> 
> My program just pushes the number of seconds forward.  Then Time should 
> answer my questions based on the new count, taking into account things 
> like Daylight Savings Time, Leap Year, etc., right?

It does, but saving time breaks the continuity. If x seconds from the epoch
means 15:00:00, x+1 seconds could mean 16:00:01.

Look:

irb(main):001:0> t=Time.mktime 2004,"mar",27,23,30,0
=> Sat Mar 27 23:30:00 CET 2004
irb(main):002:0> t+86400
=> Mon Mar 29 00:30:00 CEST 2004