I was investigating some bug that appeared after switching from 4.1.3 to 4.2 regarding Tree structures, and I've found this change in the source:

4.1.3
Code:
    registerNode : function(node, includeChildren) {
        var me = this;

        me.nodeHash[node.getId() || node.internalId] = node;
        (...)
    }
4.2
Code:
    registerNode : function(node, includeChildren) {
        var me = this,
            children, length, i;

        me.nodeHash[node.internalId] = node;
        (...)
    },
I'm wondering what is the advantage of the second approach ? In this case finding nodes by Id's won't work anymore. What if I'm adding a node with know Id and I want to reference it after adding to the structure ? I'm not considering this a bug but rather I'm interested what was the idea behind this change.