Database Reference
In-Depth Information
determination of the initial status and the process performed by a run-to-completion step as
defi ned in the UML semantics.
The chapter is organized as follows. In the following section we compare our proposal
with other related works. This is followed with a section detailing our view of dealing with
the behavior of dynamic systems, presenting the architecture in which we base our approach.
Next, we show the metamodel of UML State Machines we propose. And, fi nally, conclusions
and plans for future work are presented.
RELATED WORK
There are several relevant aspects that must be outlined with regard to the compari-
son of the approach we propose in the present chapter with other works in the literature.
First of all, it must be noted that our proposal provides a complete formalization of State
Machines, whereas, Kwon (2000) and other works (Engels et al., 2000; Gnesi, Latella &
Massink, 1999; Latella et al., 1999) consider only some basic constructs, rendering their
proposals incomplete. Among these, it is worth comparing our approach with Engels et al.
(2000), since in this work a metamodeling approach is also proposed for representing the
dynamic semantics of State Machines. These authors propose to extend the UML State
Machines metamodel (OMG, 2003) with state information that can be viewed as providing
the metamodel with information of the Snapshot Layer of our proposed architecture. Apart
from this shared issue, the remaining aspects are quite different mainly because they adopt a
different way of metamodeling the run-to-completion step: they use collaboration diagrams
and we use the notion of map. Reggio (2002) also proposes a metamodeling approach which
can be interpreted according to the architecture we present. On the one hand, in Reggio
(2002), Labelled Transition Systems are used as the semantic domain of state machines, by
means of which aspects related with the Snapshot Layer are captured. On the other hand,
they propose the use of Labelled Transition Diagrams in order to represent the change of
state within the statechart, instead of mappings as in our proposal.
The notion of map or transformation has been claimed by several authors (see, e.g.,
Domínguez & Zapata, 2000; Domínguez, Zapata & Escario, 2000; Marttiin, Harmsen &
Rossi, 1996; Saeki, 2002; Hofstede & Verhoef, 1997; Verhoef, 1993) to be a necessary arti-
fact for solving similar problems within the fi eld of method engineering, problems such as
method interoperability or method adaptation. This necessity has been recently recognized
by the UML community with the advent of the Model Driven Architecture, MDA (Miller
& Mukerji, 2003). Within this architecture, models are leveraged to be primary artifacts
during software development, and so are transformations. MDA is based on several OMG
standards, such as UML or MOF — the Meta Object Facility (OMG, 2002a), considered as
the meta-metamodel of the UML metamodel (OMG, 2003, pp. 2-6). In particular, the ongo-
ing process of development of the new UML 2.0 and MOF 2.0 embodies the Request For
Proposals of the MOF Query/Views/Transformations (QVT) (OMG, 2002b), also known as
'Unifi ed Transformation Language'. Other artifacts have been proposed in the literature for
specifying the sequence of steps of a procedure. For example, the value of process model-
ing for representing in a rigorous and explicit way this type of functional aspect is proved
in Song and Osterweil (1994). An objective, in-depth analysis of which artifact is the most
suitable for representing the procedures for calculating the initial status and the next status
in the context of representation of behavior remains an ongoing project.
