Re: OT: A320 safe after land gear problem


zoltan wrote:
> Microprocessors are unreliable for multiple reasons. One is the way the
> software is developed. High level languages, compilers, operating
> systems etc. distance the program developers from what really is going
> on.

That largely depends on what the system is and what it's being used
for. This is a sweeping and unreliable generalisation.

Another reason is the very small but finite chance of a bit error
> causing a jump to an invalid or incorrect address. Software tends to be
> like an incredibly long daisy chan with one instruction following the
> other. As soon as the chain is broken everything after that point in
> time is incorrect. I have developed methods to make microprocessors
> reliable. One such method is to apply a periodic reset which is
> followed by a short sequence of instructions. If something gets messed
> up, the next reset will start all computations from scratch. Another
> method is to use node and connection lists to drive the computations.
> In this way the connections are computations that deposit data into the
> nodes that are memory elements. Just like the brain. I made a four
> legged walking robot that learnt to walk with such a system.

Good for you, but I'm not entirely sure what point you are making here
with regard to the aerospace and embedded systems software industry.

> I prefer to write my own editor, development system etc.

?

Our customers prefer we use industry standard tools and work to
recognised industry standards and documentation standards.

You will find that to be the case in many industries. The civil
aviation industry, in particular, likes traceability.

> I actually like fly by wire, I just think it should be done right, so
> it is reliable. In my systems everything is done on a millisecond cycle.

The problem is doing things right on something as insanely complex as a
passenger aircraft flight control system is not going to be right on
the first go.

.

Relevant Pages

  • That Elusive $100 PC
    ... Instead of a closed system of ace developers with mega budgets ... of industry are at work – ... upgrade these products and develop the ancillary support sector more ... we saw a flourishing consumer electronics industry until ...
    (rec.arts.movies.local.indian)
  • Re: Peter van der Linden
    ... installing folks from the marketing department in my office and having them ... gerbil, and may not be so easy to train. ... line it became "ok" for developers to basically piss on their colleagues ... this industry would doubtless be in far ...
    (comp.lang.java.programmer)
  • Re: OT: A320 safe after land gear problem
    ... Microprocessors are unreliable for multiple reasons. ... systems etc. distance the program developers from what really is going ... method is to use node and connection lists to drive the computations. ...
    (sci.space.policy)
  • Re: Where is Open Source for FPGA development?
    ... many of the large open source projects are full time staffed ... while also having hat's as core developers in various distro's ... the industry. ... enough funds to support their product lines (IE dropping support for ...
    (comp.arch.fpga)
  • Re: Revolution: An Industry Divided?
    ... it came to me that maybe the industry has reached a tipping point ... could start giving token amounts of money to small developers to make ... innovative games at that. ...
    (uk.games.video.misc)
Quantcast