Re: type safety


Subject: Re: type safety
From: Macneil Shonle (Macneil.Shonle@Sun.COM)
Date: Mon Feb 04 2002 - 10:07:31 EST


> After significant pondering, I have come to the conclusion that Milner
> should have said
>
> Typed programs don't go wrong __as often__.
>
> There is still the issue of "going wrong C-style" (when the compile-time
> type checker gives you absolutely no hints about why a program may have
> crashed or why it may produce a random value). I begin to believe that
> nothing short of a compiler-correctness theorem can express this issue.
>
> -- Matthias

What's meant by "going wrong"? Does division by zero count? Taking the car of an
empty list? Running out of memory? This would seem obvious, so perhaps there's
another way to go wrong that isn't one of these issues and that does subvert the
type system?

Thanks,
Macneil Shonle



This archive was generated by hypermail 2b28 : Mon Feb 04 2002 - 10:07:34 EST