[SystemSafety] Collected stopgap measures

[Olwen Morgan]

On 16/11/2018 13:46, Derek M Jones wrote:

<snip>

> process, specifically documented requirements or documented design. He
>
> The lesson to learn here is that successful software does not need
> basic requirements of a professional engineering design process.
>
> <snip>

A while ago, I posted here some mischievous C code fragments that 
demonstrated instances in which clang, a compiler, performed at least as 
well as cppcheck, an (alleged) static checker. As a static checker for 
C, QAC would beat cppcheck hands-down. One of the reasons for having a 
professional engineering process is to ensure that you build the right 
thing. The developers of cppcheck, in following a dictum (dogma?) of "no 
false positives" has produced something that may be successful by its 
own lights but would, on the basis even of quite limited tests, fall 
spectacularly short of technical adequacy in a critical engineering context.

That's one reason why critical systems engineering needs defined 
processes using fit-for-purpose techniqes and tools.


Olwen