D'oh, right. A good test might be difficult. In the absence of that, I guess the safe bet is just to count every Swift binary as a false positive for ObjC, though that's not quite fair, since you can mix the two in the same code base.