When does recursion produce a StackOverflowError?

Question

What I want to know is what is the limit of recursion within a method and how is it determined? In one of the topics posted earlier in this forum, http://forum.java.sun.com/thread.jsp?forum=4&thread=231735, the answer says that the default stack depth is 512Kb or 1Mb. Is this of the method stack frame? While starting the JVM, if we use the java -Xss option, we can change the default thread stack size. Is this stack the one that is used when a method recurses? Does the number of local variables in the method or their size eat up this stack space?
I ask this question for the following:-
Before I came to know of TreeSet, I had implemented a simple binary tree and was using it to sort a collection for showing a tree in a JSP page by in traversal. Each of the tree nodes had a simple map of values. I found that when the number of elements increased to 8000, that is, 8000 nodes in the tree, StackOverFlow error started occuring. I already solved this problem using TreeSet but am curious to know why this occurs.

Answer 1

Please correct me if I am mistaken here but the amount
of data in the tree is not the direct cause of the
stack overflow.Yes, it is not a direct cause. But indirectly, it is.
The following method is in class B:-
public void invokeMethodOnInTraversal(Object objData) {
if (arrSortedList == null) arrSortedList = new ArrayList();
arrSortedList.add((SSTreeEntity)objData);
The following method is in the super class A of B:-
protected void inTraverseTree(TreeNode treeNode){
if (treeNode == null)
return;
inTraverseTree(treeNode.getLeft());
invokeMethodOnInTraversal(treeNode.getNodeData());
inTraverseTree(treeNode.getRight());
In inTraverseTree(), if I print the number of times it is invoked, the StackOverflowError occurs at 3355.
It then follows
that the stack is filled (at least partially) with the
local variables in your method.This means that objData is filling up the stack. objData is just a map with one entry. There is only one local variable in the method. This is actually the node data of each node in the tree. So as the number of nodes in the tree increase, so does the number of stack frames and hence, the stack overflow.
But is it rare to recurse 3000 times? If not, can I increase the stack limit in some way? From what I found from this link, http://www.artima.com/insidejvm/ed2/jvm8.html, where the stack is allocated is implementation dependent. One implementation may allocate stack frame on the heap, for which increasing the max heap size should work. Another may not, then what?