[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: Y2K

> >Lately I have been hearing lots of subtle and scarey ways that the Y2K bugs 
> >might affect a whole range of things.  Some American cars have been
> identified 
> >as having Y2K problems- Sorry, I don't have the specifics, this came up as
> an 
> >example during a Year 2000 briefing.
> >

> Well, unless the component knows the YEAR, there's no Y2K problem. (Other

> Saw an 89 Eldorado that had a date and year display - betcha that won't
> work on 1-1-2000! But the car should still run just fine.

Well, if its just a digital clock with a 2-digit year, it will 
roll over to 00 and work, or maybe they designed it with Y2K in
mind anyway, and it will work for longer - my simple old
Casio calculator watch, designed in the early 80s, fails after 2079
according to the manual. If this Eldo uses a pure binary 
representation, teh display may not know
what to do with the binary '100'.

What kills me with all these scare reports, besides the
non-mention of all Unix flavors, which use a counter
that rolls over in 2038, is the fact that they always talk about
software using a 2-digit representation only. While many program
only display 2 decimal digits, few devices other than plain
digital clocks with a (decimal! - actually BCD, 4 bits that make
one decimal digit) counter for each decimal *display* digit
actually use separate decimal digits in the underlying software.
The stereotypical 2-digits software would use "years from 1900"
(or, for another example, years from 1980), which would be
printed/displayed/whatever as a 19 (if that's actualy shown)
followed by the number, e.g. 19 followed by 98 (or, for the other
example , 1980+number, so its 1980+18= 1998, or 19 followed
by 80+number).
When the rollover occurs, we get misprints: depending on the 
complexity of teh print routine, we get 1900 or 19100 or,
less likely, 19 0 or 190 (or, again, possibly 2000, or 19100).

Actual date calculation which are not repeated day-by-day
and thus date-independent would be done based on duration,
i.e. the difference between year numbers: some policy
vests 15 years after being startet in 1985? No problem,
100-85=15, bingo.

I'm not saying there is no problem - I, too, will have a little
more cash than usual in hand around the time, but it seesm 
that the overblowing is going a tad far.   
Sorry for the non-Audi BW, but it just hurts - TV news
yesterday: "...but in Rhode Island, people are worried."
Followed by interview of *one* scaremonger, that sounded like
he makes his living this way.

We may have teh larget problem form people emptying grocery stores
beforehand, and everybody withdrawing all their savings a
the same time (How' *that* for the end of civilisation?)