In C++, there are two forms of binary operator overloading you can use when designing an API. The first form is to overload the operator as a member function of the class, and the second form is to overload the operator as a friend function of the class. I want to explore why you would use one form of overloading instead of the other, using a Fraction class as an example.
