Design Automation : Making Formal Methods

y Formal methods are easily dismissed as heavy weight and irrelevant. People legitimately ask to see what impact formal methods have had on software development practices outside of those areas where governments or regulatory agencies have mandated their use. We argue that design automation is an opportunity for the insertion of formal methods into software development practice. In particular, we advocate the development of small, domain-speciic design languages, and the use of these languages as front-ends to program generation systems. Software Design for Reliability and Reuse (SDRR) is a method developed at the Paciic Software Research Center to support the development of generators for domain-speciic languages 1]. Our method, which utilizes a robust suite of reusable transformation tools, has been used to implement a software component generator for a message translation and validation problem domain identiied by the Air Force. Independent contractors have used our system as part of an experiment comparing our generation-based technology to an existing solution that uses a program templates based reuse technologyy16, 18]. Preliminary results show a productivity improvement factor of 2.9 over templates, which already represents a signiicant productivity improvement over ad hoc methodss12]. The formal methods community is currently in a phase of critical self-evaluation. At the recent 1994 Monterey workshop on formal methods, the emerging consensus was that there needed to be a new, more realistic vision for future research. Participants universally concurred that formal methods are critical to the emergence of software engineering as a well-organized discipline based, as are other engineering disciplines, on sound and well-tested mathematical models. To this end, future research must be focused on approaches that would have a direct positive impact on large-scale software development. Crucial to this eeort are the following goals Incorporate formal models and algorithms in computer tools designed to solve problems of practical signiicance Recognize that practical software development is largely driven by user requirements changes Authors supported in part by Air Force Materiel Command.