> 'BSD retains this functionality for ints, but for quads the one hilited
> above different (gives -2). Does this mean that the NetBSD moddi3
> function should not return negative numbers?

I don't know what moddi3 does.

However, it seems quite logical that:

	FOO x, y, d, r;

	x = M;
	y = N;
	d = x / y;
	r = x % y;

should yield the same results for "#define FOO int" as for "#define FOO
long long", assuming M and N are in range, etc.  (That, of course, is
modulo any wacked out C promotions happening, but I don't think those
would be a problem in the example you've described.)

The fact that you're getting a different result for that case makes
_me_ think that that the code giving you that result is buggy.


chris