BP> struct base {
 BP>     int f();
 BP>     int f(int,int);
 BP> } ;
 BP> struct derived : public base {
 BP>     int f();
 BP> };
 BP> derived a ;
 BP> a.f(1,2) ;                // ERROR
 BP> [Redefining] f() and inheriting parent's f(int,int) without 
 BP> additional problems would be [more] intuitive, and make no more 
 BP> work to the compiler.
Yes it would.  It would complicate name lookup enormously, since it would 
have to perform overload resolution across multiple scopes simultaneously.  
As the current language rules stand, name lookup occurs first, finding the 
`f' in the scope of class derived, and *then* overload resolution occurs, in 
the scope of the name that is found.
 ¯ JdeBP ®
--- FleetStreet 1.19 NR
---------------