The Specification of Dynamic Distributed Component Systems

Modern computing systems are terribly complicated - so complex that most system designers and developers can only hope to understand their small piece of the larger project. The primary technologies that help system builders manage this complexity are object-oriented and/or component-centric, and the primary tools are those that assist in system modeling and specification. It is my belief that the next stage in managing system complexity comes in the form of system specification through formal methods. Only with precise, complete, and consistent descriptions of our systems and their components can we hope to understand the hyper-complex engineering that has become prevalent in computing today. But, only through the introduction of some middle-ground, semi-formal technique can modeling and specification break through into the mainstream. Such a specification methodology can''t be too hard to use, but need to be formal enough that it will help system designers and tools check the consistency and completeness of the system and its components. This thesis is the first step on the road toward formal specification of dynamic, emergent, distributed component systems, and addresses all of the requirements mentioned above. I introduce DESML: a set of new modeling constructs which can be used as a thin layer on top of most modeling languages. DESML is a variant of the Unified Modeling Language (UML), not an extension. I have redefined the the core metamodel, thus the new language is no longer compatible at the meta-level with UML. Note that such a modification is not necessary it is only a convenience in the definition of our new language. The reader should be familiar with the Unified Modeling Language and at least one formal specification language. Suggested references include [42] and [50, Chapter 2] for UML, and [20, Chapter 6] for a specification language (in this case, Z [152]).