“-—not a functional security issue”-—WRONG-—in spades. If one of my critical systems lets go in an unpredictable, untimely, catastrophic way I dam sure want to go eyeball-to-eyeball with the retrograde bleephole who made it AND the slimeball who sold it to me . If same bleephole is hiding in China, India Mexico etc he’s out of my reach. You go ahead and bench test all the stuff you can get your hands until you’re happy. But you’re not going to do, oh, say, a million cycle reliability test on your Heathkit Hobby Set. And my suppliers all know I got zero sense of humor. They’d better not sound like Mell Tillis ordering breakfast if something goes south. And we’re usually talking big bucks on both the hardware and consequences of failure side.
You want to test your way into quality assurance then YOU drive it.
“You want to test your way into quality assurance then YOU drive it.”
Nail, meet hammer!
This is the problem - people think you can test in quality. That just simply cannot be done. You can test functionality at a point in time on the manufacturing floor after you’ve done certain things to mete out infant mortals; all the while doing your level best to provide a test with adequate coverage, yet you cannot ever guarantee 100% that what you send out the door will be reliable.
You can sometimes get to 98-99% yields and see six sigma like performance in the field if you take the time to first sample properly (many fab houses charge you dearly for adequate sampling), do things like outlier elimination using a proper test and meaningful statistical data, not following logical fallacies and using a sound philosophy in the first place, but you never can cover 100% of all scenarios.
I like the way you think - too few of us nowadays!