Eviss: a visual system having a spatial parser generator

A visual system called Eviss is proposed. Visual systems proposed so far have been fixed on certain specifications. It is hard to make and modify such visual systems. Since Eviss has a spatial parser generator based on CMGs, it can construct various visual systems if provided with grammars which consist of production rules. In Eviss, a person who implements a visual system first defines production rules using figures. Eviss extracts constraints which hold between the figure elements. Then the person modifies them and write actions, if needed. An action is described as an arbitrary Tcl/Tk script. We can perform actions, such as calculating values, creating widgets and redrawing figures. Given a certain grammar, Eviss can parse a visual sentence (a collection of figure elements) spatially. Parsing means to apply production rules to the figure elements. When a production rule is applied, Eviss redraws the figure elements so that the constraints between them always hold. Actions, if any, are also executed when the production rule is applied. We also describe two examples of making visual systems with Eviss, i.e., the GUI creator and the subset of YISPATCH (Y. Harada et al., 1997). Eviss and the examples described in the paper are fully implemented and can be obtained via WWW