lecturer: Nicolas F. Rouquette

The Executable UML specification became an OMG Technology Recommendation in September 2008 and finalization is expected for September 2009. Before discussing practical applications of this specification, this talk will introduce the two key objectives of this specification:

1. enabling a chain of tools that support the construction, verification, translation, and execution of computationally complete executable models based on a foundational subset of UML; and
2. providing a general and standardized facility for specifying the structural and behavioral semantics of MOF-based modeling languages, including extensions of the foundational subset such as UML state machines.

The notion of computational completeness unifying these objectives has had a profound influence on the organization of the Executable UML specification. Whereas the 39 leaf packages in the UML Superstructure are organized according to layered package merge/import increments and dependencies, the 9 leaf packages in the Executable UML Abstract Syntax define complete syntactic units and defer package merge/import increment/dependency layering to the compliance levels. Albeit unconventional, this organization simplifies the conceptual package-level correspondences between abstract syntax and operational semantics and the relationships between operational semantics and the underlying axiomatic formalization.

The second part of the talk will focus on the normative mapping from Java to UML Activity diagrams. Tool support for this mapping enables the reverse engineering of an hybrid Executable UML/Java model into a pure Executable UML model. A hybrid model defining an interpreter for a language is called computationally complete when it, as a hybrid model, can execute itself, as a reverse engineered pure Executable UML model.