in reply to
Re^6: OT: Mathematics for programming (again)
in thread OT: Mathematics for programming (again)
( Hmm, this formulation leads me to believe that writing a proof with a computer is equivalent to the halting problem. But don't ask me to prove it. )
I suspect that would be related to the fact that there are statements which are true, but cannot be proven in a formal logic system.
The question is, is that set of errors disjoint from the set of errors that we are already good at catching? Or more simply, is automated validation useful?
A refinement of that last question: what is the cost/benefit ratio of the extra validation?
We can be fairly sure that most of the high value / low cost checks have been done already. Syntax checking (highlighting), warnings, strict and taint.
Perfect verification is impossible. We can use our computers to improve our ability to catch a certain (unspecified) set of errors in our mathematical reasoning.
This brings to mind another limitation... computers can Do What I Say. Perl's built in flexibility can Do What I Mean, most of the time, when I say things in an unusual way. But there is nothing a computer can do if I do not Say What I Want... Even humans have trouble with that.
How do you verify your meta-code which tells the verifier what you really wanted to do? Then play the $work game, where what you think is wanted is not exactly what the $customer wants, and what the $customer wants is not exactly what the $customer wants. :)