As he says, it makes the (compiled) code smaller, which often means faster since more stuff fits in the instruction cache. Though in this case I guess he could just have it be the last thing in the function since both branches go there anyway. (Edit: nah, I thought of it wrong -- it can't just go in the end.)