I am testing a Web-based application developed using ASP/HTML with ActiveX controls. There is a tree control (Sheridan Tree View). I use Robot to learn it but it seems that Robot only can understand the tree control itself but cannot recognize the nodes and leaves inside the tree control. As a result, all my mouse actions to navigate/activate a specific node are recorded with position-based (i.e. click on (x,y)). This leads to a problem that if next time I run the script and the node item is not the same node as when Robot recorded, the operation will be failed. For example, at first time clicking on (100,210) will click on node "A" but nex time clicking on (100,210) may click on node "B" because node "A" was removed.

Does anybody know any way to make Robot "learn" the sub-element of the tree control such as nodes and leaves? Please help me.

Thanks a lot.