The meta-modelling facility (MMF) consists of a meta-modelling language (MML), a meta-modelling tool (MMT) and meta-modelling method (MMM). It was developed (and is continuing to be developed) to support the precise definition of families of modelling languages. The MMF was first proposed as part of a feasibility study carried out for IBM, which showed how UML could be rearchitected as a family of languages. This work is currently influencing the UML 2.0 effort, with support from Rational.

MMF papers

Feasibility Study in Rearchitecting UML as a Family of Languages using a Precise OO Meta-Modeling Approach  S. Brodsky, T. Clark, S. Cook, A. Evans, S. Kent (9/00)

• The MMF Approach to Engineering Object-Oriented Design Languages T. Clark, A. Evans, S. Kent, P. Sammut (12/00)

• MML and the Metamodel Architecture J. Álvarez, A. Evans, P. Sammut (1/01)

• An Action Semantics for MML J. Álvarez, T. Clark, A. Evans, P. Sammut (3/01)

• Mapping between Levels in the Metamodel Architecture J. Álvarez, A. Evans, P. Sammut (3/01)

• Engineering Modelling Languages: A Precise Meta-Modelling Approach  A. Clark, A. Evans, S. Kent (4/01)

• A Proposal of a Dynamic Core for UML Metamodelling with MML G. Reggio, E. Astesiano (4/01)

• Unification of Static and Dynamic Semantics of UML A. Kleppe, J. Warmer (7/01)

UML 2.0 Feasibility Study

With the support and collaboration of IBM we have conducted a feasibility study into rearchitecting the definition of UML using a precise object-oriented meta-modeling approach. The foundation of the approach proposed in the report is a new kernel language called MML, the Meta-Modeling Language.  As a part of the feasibility study, a prototype (Java) tool has been developed implementing MML. This tool is free, subject to a GNU licence, and is available for download below. 

• Report ("Feasibility Study in Rearchitecting UML as a Family of Languages using a Precise OO Meta-Modeling Approach") 

• Tool (MMT)

---

Responses to the IBM Proposal

Since the above report that introduced MML was made public, the pUML group has received a number of responses that have highlighted potential issues and improvements to the MML metamodel: 

• Coherent mechanism for dealing with classifier content inheritance  A. Evans (28/9/00)
• Remarks on the MML study  A. Kleppe (10/11/00)
• Analysis of multiple classification  T. Clark (10/11/00)
• Towards a dynamic core  E. Astesiano & G. Reggio (20/11/00)
• Notion of a variable expression  A. Evans (23/11/00)
• MML & MMT issues  T. Kühne (18/1/01)

All suggestions are welcome and will help to refine the metamodel.  If you have any comments regarding MML, please e-mail them to: 

puml@cs.york.ac.uk