no thanks - remote controlled passenger planes?

Larry Dighera wrote in
news
One presumes there are butt-loads of computer code involved in
implementing such automated aircraft operations; what could go wrong?
:-(

I have programmed as a hobby for at least a couple decades.

A quick story. On a small recent project I wrote some code
on a microcontroller to drive some full color LED's for an
art project. Worked wonderfully. Development continued for
some time.

Then I made a small change. Instant crash and burn. No lights.

It took me days to track down the bug. I ended up having to
'manually' run the code on paper and track what every variable,
every memory location, every CPU register was doing at every
step.

Turned out it was not the recent change I had just made, but
a bug that I inadvertently made very early in writing the code.
It just coincidentally happened to still work correctly until
the later change made it not.

Oh, and this code was only 33 lines long!!!

The error? I forgot to put in a "#" symbol. The result was the
command was accessing the wrong memory address to get some data,
but by pure coincidence that wrong address contained the right
data, thus it still worked. When I made the later change, it
changed the value in that 'wrong' memory address thus exposing
the error.

It amazes me to no end that people can write software for such
critical systems as nuclear power plants, aircraft autopilots,
ICBM's, and Mars rovers and NOT have the thing just end up a
heaping pile of goo. Well, even as 'just a hobbyist' I do
understand how, but it's still amazing.

But, sometimes it does still blow up in the end...

http://www.around.com/ariane.html

Brian
--
http://www.earthwaves.org/forum/index.php - Earth Sciences discussion
http://www.skywise711.com - Lasers, Seismology, Astronomy, Skepticism
Sed quis custodiet ipsos Custodes?