[SystemSafety] Fault, Failure and Reliability Again (short)

[jean-louis Boulanger]

2015-03-03 11:03 GMT+01:00 Nick Tudor <njt at tudorassoc.com>:

> Hi Peter
>
> Tis I
>
> The fault with the logic in your blog is that the design of your system
> fails to meet the specification; this I hope is obvious. The software is
> therefore as you suggest 100% reliable. Or not if it hits the one fault.
>
> The term reliability in systems has been hijacked to mean something else
> in software and is reinterpreted very badly to say that it therefore has a
> reliability of one in a thousand ( or whatever). Clearly if the software
> never encounters 20 it never gives an incorrect answer.
>
> Reliability models for software is still not recognised in DO-178C and
> this means it has not been recognised for over 25 years.
>
>
>
The same spirit for railway domain, no reliability model for software
another point, the probability for software failure is 1, yes the software
contain many bug and ...
it's why we used the DAL or SSIL, I prefer DAL "design assurance Level" ...
SSIl  is related to to confidence you have in the software and in railway,
the confidence is based on independant assessment (mandatory in railway).

-- 
Mr Jean-louis Boulanger
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.techfak.uni-bielefeld.de/mailman/private/systemsafety/attachments/20150303/9d7560a7/attachment.html>