Linked list stack implementation

I have used the following code for a stack implementation. The top keeps track of the topmost node of the stack. Now since top is a data member of the node function, each node created will have a top member, which ideally we wouldn't want.

• Is this good approach to coding?

• Will making top as static make it a better coding practice?

• Should I have a global declaration of top?

#include
using namespace std;
class node
{
    int data;
    node *top;
    node *link;
    public:
    node()
    {
       top=NULL;
       link=NULL;
    }
    void push(int x)
    {
        node *n=new node;
        n->data=x;
        n->link=top;
        top=n;
        coutdatalink;
        n->link=NULL;
        coutdatadatalink;
        }
        delete n;
    }
};

int main()
{
    node stack;
    stack.push(5);
    stack.push(7);
    stack.push(9);
    stack.pop();
    stack.print();
}

Any other suggestions welcome. I have also seen codes where there are two classes, where the second one has the top member. What about this?

-
In the pop and the print function, the new node seems unnecessary, since the next line modifies the pointer to point to the top. It could be simply

node *n = top;

-
The delete n; also looks superfluous in the print since n will always be NULL there.

-
If I'm right making top as static or global would result that you can't have more than one stack in an application. Linked lists usually have two classes: one is a Node which stores a reference to the next node and the data, and the second is a LinkedList class which stores the head and the implementation of the push, pop, etc. methods. You can see an example in this answer.

-
Take care of the boundary cases. Calling pop on an empty stack gives a segmentation fault error. I suppose that's not the best error handling method :)

StackExchange Code Review Q#13226, answer score: 3