Originally Posted by
HeavyMetallist
That has nothing whatsoever to do with how safety-critical software is written for applications such as airliner flight control software, let alone how it's reviewed, approved and certificated. For all the whining here from (apparently) pilots about how many posters are clearly not pilots and therefore haven't a clue about flying these aircraft, it's equally clear that very few of those supposed pilots have a a clue about how their aircraft are designed and engineered. It isn't at the whim of some juvenile nerd, it is thoroughly reviewed and questioned by older and wiser heads at every stage of the design and development process, and there is very direct (and forceful) involvement of professional pilots at every stage, starting from the original design concepts.
I have worked on such safety systems with really professional programmers and system designers it would take around 18 months to approve a design in the software through the control process. Complete regression testing was carried out every month on monthly 'builds' and after a year of updates the system went for another year of extensive testing before release for operational use. Quite often within hours faults were found by users.
There is no such thing as fault free software. Some software especially the FGMC software is built to default to fail, or gracefully degrade, rather than be 'clever' as the number of variables and unknowns the system has to work with are too high for safe coding. There will always be these times. Remember, the system has to be built to cope with not all likely inputs, not even all potential inputs but all possible inputs - and that starts to get extremely difficult and costly. It is easier and usually safer and more effective, for the system to pass the bag of bolts to the flight crew.