[MLton] Re: [Haskell-cafe] fastest Fibonacci numbers in the West

[Matthew Fluet]

> It could well be that you are right that it is in the conversion to base 10,
> but I don't see how your example proves it.  The problem is that the heap
> could be getting corrupt, and that can cause a seg-fault arbitrarily later
> or not at all.

True.  However, I also modified IntInf_toString in IntInf.c to printf at 
the start and finish of IntInf_toString, gdb stepping through 
IntInf_toString verifies that we never return from the mpz_get_str call.