Panels

Panel 1
What would be the ideal meta-modeling infrastructure?
Chair: Pierre-Alain Muller

Participants

Thomas Kühne (TU Darmstadt)
Stuart Kent (Microsoft)
Jean-Marc Jézéquel (IRISA)

Abstract

The goal of this panel is to discuss the requirements for an ideal meta-modeling architecture. The panel will address the following points:

• What is the scope of meta-modeling?
• How to reconcile domain-oriented meta-hierarchies with language-oriented meta-hierarchies?
• What would be a minimal core meta-language?
• What kind of tool support should be available for meta-modeling?

Panel 2
A DSL or UML Profile. Which would you use?
Chair: Stuart Kent

Participants

Steve Cook (Microsoft)
Bran Selic (IBM Rational)
Juha-Pekka Tolvanen (MetaCase)
Peter Graff (InferData)

Abstract

In implementing model driven approaches to software development, there is some debate about the languages to use for modelling. On one side, there is the UML, advocates of which might argue that it has everything you'll need and its profiling mechanisms are quite adequate to cope with any customization or specialization you might need. On the other side are those advocating domain specific languages, who might argue that most of the time you're going to need to specialize and customize the language you use for modelling, that UML is not a good starting point for such specialization, and UML profiles are weak mechanism for extensibility anyway. Let's instead put together technology appropriate for building DSLs, they'd say. Of course, within the DSL camp, there's further debate about what is the right technology, but perhaps that's the topic of another panel... we'll see.

Panel 3
Building Better Systems: Modeling, Verification, and Testing
Chair: Clay Williams

Participants

Paul Baker (Motorola Research)
Lionel Briand (Carleton University)
Sudipto Ghosh (Colorado State University)
Colin Campbell (Microsoft Research)

Abstract

A stated goal of model-based software development is that software quality will drastically improve as a result of the use of modeling methods. This hope has been held out repeatedly in the past by other software development movements, including formal methods, computer aided software engineering (CASE), and various process movements, such as the Cleanroom approach to building software. The participants in this panel will explore whether we are reasonable in hoping that modeling as we know it today will significantly assist with quality issues. In doing so, they will discuss what the major technical issues are that need to be addressed in order to achieve higher quality software, and propose a research agenda for addressing these issues. The panelists will pay particular attention to the use of modeling languages to facilitate better testing, as well as how modeling languages can be used as a basis for verification approaches such as model checking and theorem proving.