Question

In C++, I have two block of codes like this:

Base *base = new Base();
base->showName();

And:

Base base;
base.showName();

I don’t know when do we use pointer and when not? And what’s different and what is better?

Answer 1

The first code you showed is a memory leak.

The second snippet is Java, not C++. The question has been edited to use my suggested code.

Generally though, in C++ you should avoid new unless you really NEED dynamic lifetime. Instead, write:

Base base;
base.showName();

This is better because

• It’s faster, no heap allocation needed.
• You don’t have to remember to free the memory, the compiler does that automatically at the end of the scope.
• It’s automatically exception safe. The destructor will be called, and the memory will be recovered, during stack unwinding.

If the object needs to live past the end of the scope, you should be using:

unique_ptr<Base> base(new Base());
base->showName();

Now unique_ptr will free the memory for you when the unique_ptr dies, and it’s also exception-safe. When you return a unique_ptr, ownership is transferred to the caller, and he can reap the benefits of automatic cleanup.