There are two computers in my office at work.  One is a
12-year-old IBM P.C./XT equipped with an AST clock board.  The other
is a Compaq486/33.  The XT, believe it or not, rolls over perfectly
from Dec. 31, 1999 to Jan. 1, 2000 with no problem at all while the
CP486 which was built in 1993 gets very confused and goes to something
like January 4, 1980.  It doesn't quite start at the beginning again.
The problem appears to be in the CMOS counter because I can manually
set it to a date in 2000 and it works fine.  Basically, everything
somebody has mentioned in this thread from the individual software
applications being run to the system BIOS to the battery-run clock can
be a source of trouble if it wasn't designed without shortcuts.  I
just thought it was amusing that the old XT dinosaur did it right.  I
use that box as a terminal in to a UNIX system so it doesn't matter,
but it was an interesting test.  Of course, an XT with a different
peripheral clock module is a hole different case and it all depends on
what count the BIOS and DOS get to see when the system is powered up.

Martin McCormick