Difference between shared_dynamic_cast and dynamic_pointer_cast

Question

Can someone explain to me the difference between:

shared_dynamic_cast and dynamic_pointer_cast from the Boost library?

It appears to me that they may be equivalent.

Answer 1

Given a shared_ptr<T>, the two functions are indeed equivalent.

The difference is that shared_dynamic_cast only works with shared_ptr<>'s, while dynamic_pointer_cast works with any kind of pointer (via overloading). This enables you to perform a dynamic cast on any pointer concept, regardless of how that pointer is actually composed:

#include <boost/pointer_cast.hpp>
#include <boost/shared_ptr.hpp>

struct foo {};
struct bar : foo { void f(){} };

template <typename Ptr>
void test(Ptr ptr)
{
    boost::dynamic_pointer_cast<bar>(ptr)->f();
}

int main()
{
    bar b;
    foo* pf = &b;

    std::shared_ptr<foo> spf(new bar());

    test(pf); // turns into dynamic_cast<bar*>(pf)->f();
    test(spf); // turns into shared_dynamic_cast<bar>(pf)->f();
}

Because dynamic_pointer_cast has the capability of shared_dynamic_cast and more, the latter function is deprecated. (Likewise in C++11, there only exists dynamic_pointer_cast.)

(The idea is the same for the other cast variants too, of course.)