SHF: Small: Change Theory for Variation-Aware Programming

Oregon State University, Corvallis OR

Investigators

Abstract

The management of changes is a difficult and error-prone process in software maintenance and many other fields. Due to a lack of theory, there are few tools and methods available to systematically change objects while preserving important properties along with such changes. This problem is addressed by the theory of structured change. The overall research objective is to find principles for sound change management and to establish a theoretical foundation for the development of supporting tools. One particular goal is the development of domain-specific languages, which users can employ to effectively manage changes in all kinds of software applications. The following technical approach is pursued. First, a flexible change representation is developed. Based on this representation, laws of a change algebra will be established to identify property-preserving transformations that can support sound change transitions. This work is accompanied by the development of algorithms for the incremental checking of property preservation. Ultimately, the theory will enable the development of tools that help users to effectively manage the evolution of objects. This goal is supported by the investigation of interaction principles that underly the editing and exploration of structured objects and their changes. Building on top of the theoretical foundation, the development of domain-specific languages will provide concrete help for users to express more sophisticated transformations and combinations than the simple one-step operations that are offered by the underlying formal model. Specifically, in the area of change representations for programs such a DSL will yield new, theoretically founded support for feature-oriented programming and software product lines. In the context of spreadsheets, web sites, etc. the development of such DSLs will empower millions of users to deal in a more systematic way with changes.