I can't resist !
Firstly, the date (seconds since 1904) is stored in 32 bits which is a
long. Currently most
processors including the draggonball do 32 bit (max) integer style
arithmetic.
Therefore, it is not just adding a couple of bits, it means adding
another 32
bits. However, all the arithmetic functions now increase in complexity
since
1 instruction to add a value (assume only 32 bits) becomes 3
instructions
(everywhere !) i.e. add value to 32 bits, then check carry flag and if
set
increment upper 32 bits. It is even more complex if you add a 64 bit
value.
The next generation microprocessors (IA64 and the like) will include 64
bit arithmetic but don't expect to use one soon - there too power hungry
for now.
When we get to use a 64 bit cpu the time could change to 64 bits and all
the API's would still work the same as now (i.e. still offset from
1904), so the only change would be anyone accessing the timeinseconds
(and if you're using C you probably will only need to recompile (which
you would have to do anyway since it's a new cpu !
Just my 2c
Ray
