Nodes - Unique Identification

[nickman]

I am in the process of documenting some of the primary structures in the production profiler and their constituent cardinalities.

One issue I have run into is the cardinality and identification of Nodes. I had originally thought that a Node would be unqiuely identified as a Category Name and an Operation Name.

For example, consider an EJB Method "OrderEJB.processOrder". We might assign it an identifier as follows:

Category Name: Stateless Session Bean
Operation Name: OrderEJB.processOrder
Unique Identifer: "Stateless Session Bean:OrderEJB.processOrder"

However, what this does not take into account is the potential for different execution paths within that operation.

OrderEJB.processOrder-->Op A --> Op B --> Op C
OrderEJB.processOrder-->Op C --> Op X --> Op Y

In this instance, I would think these two code paths should be represented by two different nodes, but the naming and tracking will now get substantially more complex.

Any thoguhts ?

//Nicholas

[clebert.suconic]

The way I though it should work is:

OrderEJB.processOrder is a node.
OP A, is another node, under OrderEJB.processOrder node.

We will have to navigate in trees. We shouldn't have a flat structure.
Because you still need to analyze the root node processOrder combining A + C = total of the node.

Makes sense?