Applied Informatics

After years of dominance, classic desktop-based WIMP (Windows, Icons, Menus, Pointer) systems are slowly being replaced by modern post-WIMP systems. Such systems do not stick to a certain user interface or interaction paradigm, but rather contain a heterogeneous set of characteristics that stem from multiple fields of research. These characteristics induce a variety of different challenges that designers and developers of post-WIMP systems have to face and overcome. In this thesis, it is argued that formal methods, and in particular finite-state machines, are an important means to tackle certain of these challenges. In order to adapt finite-state machines to the requirements of post-WIMP systems and to improve their expressivity, specific additions to their default notation are suggested. One such addition allows the specification of animated transitions. The other addition is a notation to differentiate multiple input points. Although there is a consensus between many researchers and developers that finite-state machines are a rather natural formalism for the specification of complex interactive systems, their implementation is not yet supported appropriately by user interface toolkits or programming languages. Thus, a finite-state machine framework for post-WIMP interaction design, the Reactive State Machine framework, is presented in the main part of this thesis. Due to its declarative nature, it greatly facilitates the transformation of a graphical state machine model into code. The Reactive State Machine framework supports all important state machine concepts, such as states and transitions. What sets it apart from other similar frameworks is its full support for input events, its support for animated transitions and its support for the multi-point notation that is introduced in this thesis. To show the utility and value of the Reactive State Machine Framework its application in three assorted use cases is demonstrated. For one of the use cases, the Facet-Streams system, a comparison is conducted between the old naive implementation based on low-level implementation techniques and the revised implementation based on the Reactive State Machine framework. Finally, the threshold and ceiling of the Reactive State Machine framework are assessed in a brief informal evaluation. To conclude the thesis, its main contributions are summarized and an outlook on potential future work is given.