MODELING AND IMPLEMENTATION OF VISUAL LANGUAGE USING PETRI NET MODEL

Abstract

In this thesis, we introduce a new approach to the implementation of visual languages. Instead of the usual way of representing visual language constructs using one-dimensional textual language, we represent these constructs in terms of simple, two-dimensional graphical objects. In particular, we develop a variant of Petri net model, the Modified Predicate-Action Net, to represent a class of visual languages. Using this model, we show that the constructs of a typical dataflow-oriented visual language, the Prograph, can be easily and accurately represented by the components of our net model; the control flow and data manipulation of the Prograph are shown to be monitored by the control : 1 data aspects of Predicate-Action net theory. Furthermore, the execution of a Prograph program can be performed efficiently through the direct execution of the net model. From the Petri net point of view, the Modified Predicate-Action Net adds three notions to the current Petri net theory: the use of process ids to denote the multiple instantiations of nets and the hierarchical relationship among these instantiations; the use of tokens carrying data information to model the data-oriented aspect of the system modeled; and the introduction of functional abstractions to model the procedural invocations in a programming language. From the visual language point of view, the Modified Predicate-Action Net can be used not only as a base for implementing various dataflow-oriented visual languages, but also to support programming that involves concurrency. Lastly, some analysis can also be performed on the visual program via the analysis of Modified Predicate-Action Net; this marks the beginning of intrusion to the formal aspect of visual program verification.