Difference between these shared_ptr instantiations?

Question

I'm not so experience with templates, and I've found some usage in my team's code that has me confused.

Let's say I have a base class with the following two typedefs for shared_ptr:

class Foo
{
     virtual ~Foo();
     virtual void bar() = 0;
};

typedef boost::shared_ptr<Foo> FooPtr1;         //this is what i'm used to seeing
typedef boost::shared_ptr<class Foo> FooPtr2;   //this is in our codebase

Is there any difference between these typedef's? Is it related to usage with derived classes?

Answer 1

There's no difference.

The syntax class Foo is supported for consistency with struct Foo, which in turn is supported for C compatibility. In C a struct is not a type by itself: the type corresponding to a struct called S is struct S, which is usually named via a typedef. In C++ the typedef isn't needed, and the syntax is largely irrelevant. I've found it useful for introducing "inline" tag types for template instantiations, like X< whatever, struct X_tag >, where the tag type is an incomplete type. But that's all.