> No, I meant exactly what I typed.   Read the standard.  Unless I'm
> completely misremembering, it is indeed permissible for sizeof (int)
> to be 64 and sizeof (long) to be 32.

I don't have the standard handy, but I'll bet section 3.1.2.5 (Types)
says that is false.  Also see 3.2.1 ([Conversions of] Arithmetic Operands);
the rationale document (http://www.lysator.liu.se:7500/c/rat/c2.html#3-2-1)
certainly seems to believe the types form a strictly nondecreasing hierarchy.