One of the main reason sequence diagrams are useful is that they can describe asynchronous activities in a generally deterministic manner.
For example, let's say we have an object (it could be a servlet, a message-driven bean, a web-service listener, a UI component, or whatever). The code instantiates this object and links it to some form of input so that a message will be called when something happens. All that the code (and thus a reverse engineering tool) can possibly express is that this method might be called, zero or more times, some time in the future.
A well-crafted sequence diagram, on the other hand, can xepress domain information such as the events, situations and scenarios which will trigger this method to be called, including when and how often it should be called in the real system.
This type of "callback" behaviour is very common in Java systems.
I don't know if this is helpful to you... I just started using JDeveloper and, apparently, it can create sequence diagrams through reverse engineering. Maybe you can create the diagram with JDev and then export the XMI to use in Rational Rose (I assume RR uses XMI).
I haven't tried it, so I don't know for sure if this will work. In fact, I'm just learning to use UML tools, so I'm not an expert. Here is an excerpt from JDev's help files...
To create a sequence diagram using the debugger:
+Open the source file containing the code from which you want to create a sequence diagram.The source code must contain at least one breakpoint.
+Select Debug-->Debug with Diagram [...].A new sequence diagram is opened.
+Step through the code using the debugger controls.The sequence model is populated with elements as you step through the code.
permaculture is a more symbiotic relationship with nature so I can be even lazier. Read tiny ad:
Free, earth friendly heat - from the CodeRanch trailboss