As for your remark about saving $0.17 on a $0.20 component, well, isn't that a worthwhile effort?
I know of no '51 derivative that sells for $0.20 and you, constantly, refere to previsous statements and this thread IS about testing '51s.

I remember that you once said that you didn't need to test LCD's beyond the functions that you use in your application. Well, that's probably true if all you care about is "getting them out the door" since that will suffice for a demonstration. If several functions fail, however, isn't the entire device broken? Would you ship a defective device in your product? If some of the bits don't fail until it's had a chance to warm up, aren't you concerned it will have some effect? Would you ship a product when the carry doesn't work just because you don't use it in the current code? What if the next field-fix requires it?
apples and oranges you refer to the carry in a LCD (which I do not use any of).
To the point: if a display works in my system before and after "it's had a chance to warm up" why would I spend time and money on testing whether the "storage for oddball characters you create yourself" works or not, since I do not, and never will, use it.

You have a very low opinion of distributors, is that reflected by your use of surplus you referred to a while ago? (was it LCDs?)

I have had NONE ZERO NADA problems with one component in 1,000,000 failing NO such failure have cost enough to justify extensive pre-testing; HOWEVER I have had very costly replacements because of bad designs by others. The likelyhood of a costly 'accident' being due to design is far greater than it being due to the component.

Go ahead, 'design' without adhereing to min/max temp etc and then blame the component manufacturer. This exactly happened with a 'design' done by someone using 'typical' values from a datasheet, getting by with it for a couple of years and then BOOM the 'design' failed - of course it was the new components - notthe design, the fact they were within min/max did not matter.

Erik