FFI Report, CVS Id 1.11

[Manuel M. T. Chakravarty]

Alastair David Reid <reid at cs.utah.edu> wrote,

> Marcin:
> > | > PtrDiff 
> 
> SimonPJ:
> > I must say that I rather agree with Marcin here.  Let's just use Int
> > (but not Int32!). On machines with big address spaces, Ints will be
> > big.  Keep it simple. Oh, and efficient!
> 
> For all the reasons I've used in the past of avoiding future problems
> (especially those associated with Int), I disagree.  But I didn't seem
> to make much headway last time so let me just make this suggestion:
> 
>   Let's add some asserts to the affected code so that we'll get warnings
>   when we try to compile the code on a system that breaks these assumptions.
>   Something like this:
> 
>   #if SIZEOF_INT != SIZEOF_PTR
>   #error "Pointer difference operation could overflow - please fix"
>   #endif

To be honest, by now I am prepared to leave the status quo
as it is, which, as you say, basically forces any sensible
implementation to make `Int' as big as pointers.

Cheers,
Manuel