[patch] Lno branch merge part 8 -- canonical induction variable creation

[Zdenek Dvorak]

Hello,

> Again your points "1)" and "2)" above show the confusion.  We really,
> need a way to say
> 
> 	1) doesn't overflow.
> 	2) undefined on overflow.
> 	3) overflow wraps around using 2's complement arithmetic.
> 	4) trap on overflow.

not really.  We basically just speak in different words about the same
thing:

> 1) Wrapping -- i.e. the compiler cannot assume anything about it.
  is yours 3); is there a reason why we should support a different
  wrap semantics?

> 2) Wrap undefined -- i.e. compiler can assume it does not wrap if it
>                      wants to.
  is union of yours 1) and 2).  Is there a reason why to distinguish
  between "doesn't overflow" and "undefined on overflow" -- i.e. are
  there optimizations that could benefit from postulating a different
  wrap semantics?

> 3) Trapping -- i.e. the compiler not only assumes it does not wrap,
>                but also adds check for it.
  is yours 4)

Zdenek