TITLE: The first argument to overloaded operators


PROBLEM: hartman@ulogic.UUCP (Richard M. Hartman), 14 Oct 92

I would like some explanation of the reasoning behind the rule
that an overloaded operator must take at least one parameter
of as "class" type.


RESPONSE: steve@taumet.com (Steve Clamage)
	  Vice Chair, ANSI C++ Committee, X3J16


One explanation:

Suppose we had
	int operator+(int, int);

When you write "i+j" or "k+2", the result would depend on whether a
prototype for that function were in scope.  This would be the source
of very subtle program bugs.

The requirement that overloaded operators must have a class parameter
means that an operator either has its predefined meaning or is illegal
without a prototype.  Operator "+" is not predefined for class types.

The guideline, stated in the ARM, is that C++ is extensible, not mutable.

Some operators have default semantics for classes.  Those operators
are further required to be members of a class for the above reason.
For example, assignment is predefined for class types, so a
user-defined assignment operator must be a member function.