Re: January 19, 2038

[Michal Jaegermann <michal ellpspace math ualberta ca>]

Richard Hadsell wrote:
> 
> Was it those, who found it dangerous to speculate about conditions 30
> years ahead, who decided 30 years ago that 2 digits would be fine to
> represent years? ;)

This was quite a bit earlier than 30 years ago and this was often a
concious decision.  A computer memory was then awfully expensive (do you
know what "core" really means?)  and these two digits meant real money.
I've read some analysis which claimed that that would be bigger money
that what it costs today to fix the problem.  I do not know if this is
really true but it may be.  One could ask, of course, why not to use
more effective year representation, as 8-bits may mean 128 years back and
127 forward, but decoding this representation was also an extra memory -
this time for programs.

What is more baffling that the stuff a few years old, like some variants
of BIOS for Intel boxes, shows this problem.  Original designers did
not take into account a huge inertia - mostly in programmer managers. :-)

  Michal