Recursion and the call stack

Recursion is more powerful than iteration (in fact, several languages omit explicit iteration constructs entirely). However, with that power, comes a cost: every time a procedure is called, an activation record must be allocated on the stack:

1:		add(3, 11) return 1 + add(2, 11)
2:		add(3, 11) return 1 + add(2, 11)	add(2, 11) return 1 + add(1, 11)
3:		add(3, 11) return 1 + add(2, 11)	add(2, 11) return 1 + add(1, 11)	add(1, 11) return 1 + add(0, 11)
4:		add(3, 11) return 1 + add(2, 11)	add(2, 11) return 1 + add(1, 11)	add(1, 11) return 1 + add(0, 11)	add(0, 11) return 11
5:		add(3, 11) return 1 + add(2, 11)	add(2, 11) return 1 + add(1, 11)	add(1, 11) return 1 + 11
6:		add(3, 11) return 1 + add(2, 11)	add(2, 11) return 1 + 12
7:		add(3, 11) return 1 + 13
8:		14

Often, the recursive call in a recursive function immediately precedes a return point:

Node * findInList(int value, ListNode * node) { if (node == NULL || node->element == value) return node; else return findInList(value, node->next); }

In this case, an intelligent compiler can automatically transform a recursive procedure into an iterative one; this optimization is called tail recursion elimination. Unfortunately, most C/C++ compilers do not implement this optimization.