DIYbanter - View Single Post

Edward Reid

On Sun, 03 Oct 2010 16:33:22 GMT, (Doug Miller)
wrote:
More nonsense. Nearly every bank in the world that didn't already know
about the Y2K problem became aware of it no later than the first business day
of 1970; that is, the first time they tried to write a 30-year mortgage that
terminated after 1999.

Banks have to project more than 30 years. For example, 40-year
mortgages exist (even if it's a stupid thing to do). That doesn't mean
they paid attention. In 1970, saving two bytes several times in each
record was a big deal. The mind set was still 80-column cards -- not
surprising since banks used a lot of punched card equipment priori to
computers. Were they given a clue? Yes. Did they act on it? Seldom --
as is evidenced by the huge amount of work banks had to do for Y2K
remediation.

Only in the sense that someone is capable of *imagining* ways in which they
could fail. The simple fact is that *no* device that was not date-aware had
*any* potential at all for failure.

I'm surprised by your belief that software systems are so robust. Even
if one particular subsystem is sound, it can be brought down by the
subsystems around it. Flight control software may have been sound
(AFAIK the stories about an airplane inverting when it crossed the
equator are pure urban legend), but airplanes have several
computerized systems on board, and not all are so purely involved with
avionics.

You're doing nothing more than regurgitating some of the less ridiculous scare
stories that circulated in the y2k newsgroup in 1998 and 1999.

I never read any such newsgroup and I'm regurgitating nothing. My
opinions are based on my own experience, with perhaps a bit of input
from IEEE Spectrum -- not exactly the most alarmist publication on the
planet.

Yes, and in most cases, the analysis consisted of asking "Does this device
know what day it is? Do we have to set the date as part of its startup
procedure?" If the answer is no -- as it is in the overwhelming majority of
cases -- then it won't be afffected.

If the question is answered without analyzing the code, then the
answer is usually wrong. Dates are found in code which one would
think, based on externals, would not need dates. Or which doesn't
actually need dates but has them anyway.

Realize that in general, even a minor error in a computer program can
make the entire program fail.

Only if you write brittle, poorly constructed programs.

Which describes a frightening amount of production code, and even more
so in 1990 than in 2010. I hope ...

Edward