Thread: Dynamic class variable

Originally Posted by laserlight

Is isB() really necessary? It pretty much means that the base class must know about one of its child classes. If you can get away without it, that would be better.

The only reason isB() exists is so D objects will know whether x() points to a B or C. What would be a better way to test this?

Originally Posted by laserlight

It should be an A*.

A pointer to a base class can also point to any derived classes as well without breaking things? If I'm understanding you correctly, and use A* as the data type for D::x and return type for D::x(), then the following should be possible, right?

Code:

B *b = new B();
C *c = new C();
D *d = new D();

//  point x to b
d->x ( b );

cout << d->x->some_class_B_method () << endl;

//  point x to c
d->x ( c );

cout << d->x->some_class_C_method () << endl;

Originally Posted by drrcknlsn

The only reason isB() exists is so D objects will know whether x() points to a B or C. What would be a better way to test this?

You can use dynamic_cast, but there is a slight performance cost using that.
Instead of isB() you can have something like ClassType() which returns an enum value specifying the class type.

Originally Posted by drrcknlsn

A pointer to a base class can also point to any derived classes as well without breaking things? If I'm understanding you correctly, and use A* as the data type for D::x and return type for D::x(), then the following should be possible, right?

Code:

B *b = new B();
C *c = new C();
D *d = new D();

//  point x to b
d->x ( b );

cout << d->x->some_class_B_method () << endl;

//  point x to c
d->x ( c );

cout << d->x->some_class_C_method () << endl;

No, if x is an A* pointer then it can only access A functions.

Right. I didn't think it would work like that, which is why I was asking about void *.

The only other option I can think of that satisfies my requirements is splitting D::x into two different variables, with accessors B *D::x_B() and C *D::x_C(). I didn't really want to do this, because it seems like bad design to have to test for a null pointer every time I want to use x.