Question

The following code seems to work with the compilers (clang, g++ on both, Linux and Mac OS) I tried, but is it guaranteed to always do what one would expect?

struct A {
  virtual void foo() = 0;
};

struct A2 {
  virtual void foo() = 0;
};

struct B : public A2 {
  void foo() {
    printf("test\n");
  }
};

int main() {
  B* b = new B;
  ((A*)b)->foo();
}

I realize this is bad practice and one should not do this, but does it work generally?

Answer 1

It’s not bad practice: It doesn’t work. It will do something allright. Not unlikely, crashing. It is all allowed since you are invoking unspecified behaviour. Edit You might consult your compiler technical documentation (refer to ABI) to find compiler-specific extensions that you might rely on.

Try using

static_cast<A*>(b)    // invalid static cast (compile error)
dynamic_cast<A*>(b)   // returns null pointer value (runtime)

What you are doing is effectively

• reinterpret_cast<A*>(b)

and the results are entirely your own responsibility implementation-defined.

Edit To Nawaz: relevant standards passage: § 5.2.10, clause

7. A pointer to an object can be explicitly converted to a pointer to a different object type.⁶⁹ When a
prvalue v of type “pointer to T1” is converted to the type “pointer to cv T2”, the result is static_cast<cv T2*>(static_cast<cv void*>(v)) if both T1 and T2 are standard-layout types (3.9) and the alignment
requirements of T2 are no stricter than those of T1. Converting a prvalue of type “pointer to T1” to the type “pointer to T2” (where T1 and T2 are object types and where the alignment requirements of T2 are no stricter than those of T1) and back to its original type yields the original pointer value. The result of any
other such pointer conversion is unspecified.