This paper shows how techniques devised for aspect-oriented modelling (AOM) can be used for aspect-oriented metamodelling. Metamodelling is an approach to language definition, which has been adopted by industry for defining modelling languages, in particular the Unified Modelling Language (UML), and involves the construction of an object-oriented model of the abstract syntax and, optionally, the concrete notation and semantics of the target language. As the importance of models in software development grows, so does the importance of metamodels, not least to provide the basis of implementations of modelling tools. Specifically, there is a need for complete and accurate metamodels, which have a uniform architecture and which can be organized appropriately to support the definition of families of languages. The approach described in this paper uses AOM techniques to address this need. It is illustrated by extracts from the 2U submission to the UML 2.0 RFPs issued by the Object Management Group in 2001. The paper concludes with a discussion on the customization and generation of tools from such definitions.

With the adoption of UML by the OMG and industry as the linguae-francae of visual systems modelling, one begins to ponder what will come next in this field? This tutorial brings a vision for visual modelling beyond UML. We present and consolidate radical new notations, proposed in a series of research papers and with quickly increasing adoption by industry, for the specification of complex systems in an intuitive visual, yet precise manner. The recurring theme of these notations is the upgrading of familiar diagrams into a powerful visual language. Spider diagrams considerably extend Venn-diagrams to the specification of OO-systems. Most familiar OO-concepts are translated to set theoretical terms: class into set of objects, inheritance corresponding to subset, and even Harel's statecharts interpreted as the set of objects in that state. Constraint diagrams enhance the arrow notation to describe static system invariants which cannot be described by UML class-object diagram. Reasoning rules are developed for the notation and strong completeness results are given. Finally, 3D-diagrams show how the third dimension and VRML modelling can be used for a conceptual modelling of dynamic system behaviour. Much of the tutorial will be based on a case study developed in industry, illustrating how the new notations are combined with those of UML, including OCL.Highlights include:• A crash critical overview in UML, stressing its weaknesses and strengths,• A rich visual constraint language and an insight into subtle issues that arise when defining a visual language, for applying the popular design-by-contract using a visual formalism• A discussion of diagrammatic reasoning with the notation, including completeness results• A case study• A demonstration of a graphical editor for the constraint-diagrams language• A look to the future of visual modelling, including ideas about 3D modelling notations and visual modelling tools.

The tutorial is example driven and illustrates how the new notations are combined with those of UML, including OCL. Some of the examples are drawn from industrial contexts, in particular the telecomms sector. Highlights include:

In our work with legacy information systems, we have found two prevalent anti-patterns-tight coupling and code pollution-which, if not addressed in replacement systems, could result in today's new systems simply becoming tomorrow's new legacy system. Tight coupling results from Explicit Invocation across collaborating components. Code pollution results from implicit (rather than explicit) reflection of time-ordered collaboration protocols. These anti-patterns diminish component maintainability, flexibility, and reusability. In response, we propose a synthesis of Implicit Invocation (which reduces tight coupling) and Statecharts (which reflect collaboration protocols directly). The paper describes the development of EventPorts, which realize this synthesis and thus encapsulate a novel and promising component collaboration technology.

The Object Constraint Language (OCL), which forms part of the UML 1.1. set of modelling notations is a precise, textual language for expressing constraints that cannot be shown in the standard diagrammatic notation used in UML. A semantics for OCL lays the foundation for building CASE tools that support integrity checking of whole UML models, not just the component expressed using OCL. This paper provides a semantics for OCL, at the same time providing a semantics for classes, associations, attributes and states.

Traditionally, diagrams used in software systems modelling have been two dimensional (2D). This is probably because graphical notations, such as those used in object-oriented and structured systems modelling, draw upon the topological graph metaphor, which, at its basic form, receives little benefit from three dimensional (3D) rendering. This paper presents a series of 3D graphical notations demonstrating effective use of the third dimension in modelling. This is done by e.g. connecting several graphs together, or in using the Z co-ordinate to show special kinds of edges. Each notation combines several familiar 2D diagrams, which can be reproduced from 2D projections of the 3D model. 3D models are useful even in the absence of a powerful graphical workstation: even 2D stereoscopic projections can expose more information than a plain planar diagram.

A new visual notation is proposed for precisely expressing constraints on object-oriented models, as an alternative to mathematical logic notation used in methods such as Syntropy and Catalysis. The notation is potentially intuitive, expressive, integrates well with existing visual notations, and has a clear and unambiguous semantics. It is reminiscent of informal diagrams used by mathematicians for illustrating relations, and borrows much from Venn diagrams. It may be viewed as a generalization of instance diagrams.

Recent research has advocated the use of deontic logics in requirements specification. A form of deontic action logic is described, and it is shown how it can be used to specify both temporal constraints and error recovery. The logic includes a deontic predicate of bounded obligation, and normative predicates. Bounded obligation allows bounds to be placed on the performance of actions, and can be used to specify temporal constraints. The formalism also allows the removal and extension of obligations before they must be met. The normative predicates indicate when a system has performed an action normally or abnormally, and allow the specification of corrective actions that recover from error situations.>