Cart
[Home] [By Thread] [By Date] [Recent Entries]
On Thu, 2004-12-30 at 16:33 -0500, David Megginson wrote: > > I find the airplane example interesting, since I just spent the > morning with my head stuck under an airplane cowling (burned-out > starter solenoid). Obviously, my Piper is fly-by-loose-cable rather > than fly-by-wire, but what makes airplane systems safe -- whether > they're mechanical or electronic -- is not error-free design and > implementation, but an enormous amount of redundancy. > > The trick is to make sure that there are always at least two ways to > do most things (in a cheap plane like mine; sometimes dozens, in > commercial airliners) and that they are truly independent of > each-other. I don't think that works on EFA and suchlike, particularly not in the realms of software. Provability and testing rules the day in that domain. Reliability increases with parallel hardware systems; Multiple processors similarly help to check each other, but the actual routines used rely on different approaches. -- Regards, Dave Pawson XSLT + Docbook FAQ http://www.dpawson.co.uk
|
