This paper describes a model-based approach for designing Distributed User Interfaces (DUIs), i.e., graphical user interfaces that are distributed along the following dimensions: end user, display device, computing platform, and physical environment. The three pillars of this model-based approach are: (i) a Concrete User Interface model for DUIs incorporating the distribution dimensions and expressing any DUI element in a XML-compliant format until the granularity of an individual DUI element is reached, (ii) a specification language for DUI distribution primitives that have been defined in a user interface toolkit, and (iii), a step wise method for modeling a DUI based on the concepts of a distribution graph expressing a distribution scenario that can be played, based on the distribution primitives. A distribution graph consists of a state-transition diagram whose states represent significant DUI distribution states and whose transitions are labeled by an Event-Condition-Action (ECA...
ABSTRACTThis paper describes a model-based approach for designing distributed user interfaces (DUIs), i.e. graphical user interfaces that are distributed along one or many of the following dimensions: end user, display device, computing platform, and physical environment. The three pillars of this model-based approach are: (i) a Concrete User Interface model for DUIs incorporating the distribution dimensions and able to express in a XML-compliant format any DUI element until the granularity of an individual DUI element is reached, (ii) a specification language for DUI distribution primitives that have been defined in a user interface toolkit, and (iii), a step-wise method for modeling a DUI based on the concepts of distribution graph expressing a distribution scenario that can be played namely based on the distribution primitives. A distribution graph consists of a statetransition diagram whose states represent significant distribution states of a DUI and whose transitions are labeled by an even-condition-action representation. The actions involved in this format may call any distribution primitive of the DUI toolkit. In order to exemplify this model-based approach, two simple DUIs are first designed: a DUI for the Pictionary game and a DUI for the Minesweeper game. They are then incorporated into a larger DUI game of the goose where cells may trigger the two other games.