The standard smart pointers

C++ has a relatively small zoo of smart pointers. Before looking at the set of options provided by the standard, let’s take a moment to show the problem we are trying to solve. Consider the following (deliberately incomplete) program. Do you see anything wrong with it?

class X {
   // ...
};
X *f();
void g(X *p);
void h() {
   X *p = f();
   g(p);
   delete p;
}

This is code that is legal but not something you want to see in a contemporary program. There’s just so much that can go wrong here, such as the following from a non-exhaustive list of potential problems:

• We don’t know whether g() will call delete p, leading to a second delete (on a destroyed object!) in h() afterward
• We don’t know whether g() might throw, in which case the delete p; instruction in h() will never be reached
• We don’t know whether h() should be assumed to own...