Question

I’ve heard it said several times that if you need to dynamically downcast then there may be something wrong with the design. I see it as a similar mechanism to COMs queryinterface, querying an object at run-time to see what interfaces it supports and then invoking the appropriate method(s) accordingly.

Is there anything wrong with this?

Answer 1

There’s nothing inherently wrong with it, and sometimes it’s appropriate, but it’s often poor design to write code that makes decisions based on the dynamic type of an object. If you have something like this:

void foo(Base const *base) {
  Derived1 const *derived1 = dynamic_cast<Derived1 const *>(base);
  if (derived1) {
    // Do stuff
  }

  Derived2 const *derived2 = dynamic_cast<Derived2 const *>(base);
  if (derived2) {
    // Do stuff
  }
}

you’d probably be better off putting that functionality into the derived classes as a virtual function, and letting the language’s built-in dynamic dispatch take care of figuring out which code to run. That way, if you add a Derived3 later, you don’t have to find all the places in your code where you’re checking for Derived or Derived2, and add a check for Derived3 as well. Just implement the virtual function in Derived3 and all the existing virtual function calls will work.