> Or maybe they just don't understand the nature of embedded software
> bugs. Real-time code is fraught with hazards along the lines of
> "what if this event happens between those two lines of code" and
> we are long since past the point where it can be exhaustively
> tested in any reasonable length of time.
I once had some real-time software that I wrote that passed all testing
and was in production for nearly a year before it started crashing.
It took me two months to find the problem (mostly to be able to
reproduce the "random" failure. It all involved a subtle timing issue
between two events. Two lines of code fixed the problem and it worked
flawlessly for years after that. But in my case, the worst that
happened was a hung database..
Tim Mullen
_______________________________________________
Support Team.Net http://www.team.net/donate.html
Suggested annual donation $12.96
Shop-talk mailing list
http://autox.team.net/mailman/listinfo/shop-talk
http://www.team.net/archive
|