Homework 16: Move Semantics, Value Categories, and Smart Pointers

In class, we explored what std::move does, and did some experiments to show that std::move(x) doesn't actually move x anywhere. What is the semantics of std::move (that is, what does std::move actually do)?

In the UniquePtr class we developed in class, how we changed the copy and move constructors as well as the assignment operators to guarantee that only 1 UniquePtr can point to an object at a time.

How do the move constructor and the move assignment operator provide this guarantee?

In the class, I argued that std::unique_ptr is a good first choice when needing a pointer.

1. (1 pt) What is the advantage of std::unique_ptr over a plain pointer (like T *)?

2. (2 pt) What are the two advantages of using std::unique_ptr over std::shared_ptr?

Reference counting cannot detect that certain objects are not reachable (and can be safely destroyed).

1. (2 pt) What causes this limitation (how the objects should be used for reference counting to fail to determined that they are unreachable)?

2. (1 pt) What machinery is provided by the C++ standard library to get around this limitation?