No abstract

Interactive computer applications are sometimes specified by their use cases. These specifications are often formulated in a natural language to enable domain experts, that are not familiar with formal notations, to validate their usefulness. A use case-oriented application framework facilitates manual translation of the natural language specifications into high level code, whose equivalence with the specifications is easy to establish. The purpose is to reduce the costs of both verification and coding. A previous framework of this kind achieved its high level by providing most of the graphical user interface (GUI) code. However, the automatically produced GUI was not always satisfactory. This study reports on advances achieved by more intelligent user interface construction framework. The study introduces a new kind of user interface component, called use case displayer, that enables an almost automatic generation of the user interface of the entire application. The framework was tested in a student laboratory, where it reduced the implementation effort, leaving most of the time (10 of the 15 available weeks) for requirements elicitation, specification development and validation. As expected, the designs produced had a higher level of usability than found in comparable student laboratories. At this stage the framework is useful for student laboratories. More research is required to assess its suitability for industrial use.

Laboratories for training students in software development involve a considerable coding effort, reducing the complexity of the exercises that are possible in the available lab time. In order to reduce this coding effort our laboratory employed the component framework WebSI. An application developed with WebSI is composed of application logic (AL) component and an interaction style (IS) component, containing code for producing and managing the graphical user interface (GUI). In the laboratory the students wrote only the AL component, while the instructor provided a number of readymade different IS components. If the student didn't like the produced GUI, she could simply exchange the IS component with one that produces a GUI that is more to her liking. The exchange of an IS component is possible since all IS components employ the same abstract semantic interface, which does not specify the geometric and other properties of the GUI to be produced. Changing components is fast, as no compilation is required. The time the students saved by using ready-made IS components was invested in requirement elicitation and specifications, which had a beneficial effect on the usability of the student projects.