[Coin-discuss] Compiling Osi with gcc 2.95.3

[Michael Hennebry]

On Fri, 18 Feb 2005, Lou Hafer wrote:

> Ted, John,
>
> 	I added the using declarations to suppress warnings from the Forte
> C++ compiler which complained that the more local declarations were hiding
> less local declarations. In effect, using brings the other declaration into
> scope for consideration in overload resolution. (Vague, I know. Will try to
> generate an example, but need to go and lecture just now.) Can't recall that
> I ever found a case where the code behaved incorrectly. It was strictly a
> matter of reducing compile time warnings.

A name in a derived class is supposed
to hide the same name in its base class.

class Base {
public:
    virtual bool fred(const Base &rhs) const;
} ;


class Derived : public Base {
public:
    // using Base::fred;
    virtual bool fred(const Derived &rhs) const; // hides Base::fred
} ;

Derived dl, dr;
Base    bl, br;
...
bl.fred(br);  // refers to Base::fred
dl.fred(dr);  // refers to Derived::fred
bl.fred(dr);  // should not compile without using
              // with using, refers to Base::fred
dl.fred(br);  // I'm not sure if this should compile
              // might be compiled as static_cast<Base &>(dl).fred(br)
              // or as Base(dl)->fred(br)
              // it might matter whether fred is an operator

In none of the above cases does the virtualness of fred matter.

-- 
Mike   hennebry at web.cs.ndsu.NoDak.edu
"Our gods are dead.  Ancient Klingon warriors slew them
... they were more trouble than they were worth."          --  Worf