Question

I am trying to figure out the cleanest way to implement using a boost::shared_ptr as a member of a class. Let’s say I have class A that contains a boost::shared_ptr to a class B member. The B class could be either the base type (B) or a derived type (let’s say C or D). This member would be obtained/set continuously throughout the lifetime of class A (using accessors/modifiers). B has to be polymorphic so I can’t just store it as B but rather as a reference/pointer to B. I also wanted to avoid having to manage the memory on my own as well. Let’s say it all looked like:

class A
{
public:
    const B& GetB() const { const B* b = m_pB.get(); return *b; } // Is the correct way to do this?
    void SetB(const B& b) { B* b = m_pB.get(); *b = b; } // Is the correct way to do this?

private:
    boost::shared_ptr<B> m_pB;
};

I know I’m probably doing something wrong but I wanted to provide an example of what I’m trying to accomplish. Thanks!

Answer 1

I think this program’s design needs rethinking.

The purpose of a shared pointer is that you leave the dynamically allocated resource within it so that it manages the life-cycle of that resource. If you’re changing what’s in there, then you’re responsible for managing the life of that resource after you replace the pointer in there and that’s going to lead to confusion.

You should probably just assign over the m_pB variable with a new shared pointer so the original one goes out of scope and frees the resource that was being held in the first place. So, assign over the shared pointer, not over the pointer it holds.