Configuration items

To implement an infrastructure, you need to determine which model elements must be individual configuration items, because certain usage policies apply only if that model element is a configuration item.

A configuration item is any element stored in a separate file. The project is always a separate file. In addition, in IBM® Rational® Rhapsody® you can store components, packages, classes, and diagrams (except statecharts and activity diagrams) as individual files.

It would be extreme for the entire model to be a single configuration item. In that case, only one person could update the model at any given time. The other extreme would be to make every element (every class and use case) a separate configuration item. Again, in simple systems where there are only a few dozen model elements, it would not be difficult to explicitly check out each element. However, this method where every element is a separate configuration item does not work well for larger systems. When you collaborate on larger systems, you might have to list 30, 40, 50, or more classes before you can realize a use case.

UML provides an obvious organizational unit for a configuration item–the package. A UML package is essentially a bag into which you can throw semantic model elements such as use cases, classes, objects, and diagrams. Therefore, although you might want to make packages configuration items in your source control or configuration management system, you need to decide which model elements go into one package versus another.


Feedback