>The there is the problem of bugs introduced by the code which checks for errors and/or bugs, ad infinitum... Even the simplist program assumes the compiler has no bugs. The compiler assumes that the ALU in the CPU has no bugs. The ALU and the memory bus partity logic assumes that the logic gates have no bugs. The logic gates assume the transistors are only two state. The the two state level assumes the engineer didn't use english units when he should have used metric units. :) (except voltages is the same in both units, so an anology can go only so far - which is another logical inconsistancy!)
>JLK

I can't tell if you're joking here Jerry, so I'll take it seriously.

The idea that bugs would be introduced by the code that checks for errors sounds like reaching to me. After all, all you'd have to do is have it report the bugs, not actually fix them. And once you fixed them then the revised code (all necessary) would have to be passed through the checking program again.

Let's assume you meant that the bug detection programming would itself have errors. That would mean that it either didn't properly detect the bug(s) it was designed to detect or it reported bugs that weren't, in fact, bugs. I suppose that could happen, but it's a natural that the developers of the debugging program would test it on itself and not allow its release until it proved "clean". [and test it not only on itself, of course]

The rest of your statement is immaterial to the discussion, since it pertains to hardware and not software. But to make one comment on it, it would be a pretty sad example of a system if, say, parts capable of working in 16 states (whenever they come along) were matched with gates operating on only two states.
People (users) already understand that hardware can fail. It really is much harder to 'sell' people the on the concept that software can fail when they fully understand that each unit is an exact digital duplicate of an original. Though a book is a physical thing, an analogy might be legitimate here. People expect that every copy of a book (of a given printing run) is virtually identical - they do not expect to find different words or sentences or paragraphs from one copy to another. And this is always true and known to be true and can be relied upon (yes, there may be printing errors, but people also know that it occurs in the transcription process from one physical medium (final draft) to another (the book) and people know that a printing error on page 326 of copy#1 will also appear on page 326 of copy#1073994).

As much as my glass is half empty regarding promotion/marketing of VFP by MS, it is half full as far as achieving "defect free" systems is concerned.

Cheers
JimN