Specification-Based Software Engineering with TAGS

Aiming at the cause, notjust the symptoms, a system based on a new software development paradigm promises a curefor the software crisis.ham, and Green ' asserted that the existing software paradigm (Figure la) has fundamental flaws that exacerbate the maintenance problem. The authors identified two flaws: (1) there is no technology for managing knowledge-intensive requirements analysis and design activities, and (2) maintenance is performed on code. The authors proposed a new, automation-based paradigm (Figure lb), for which "the [support-ing] technology... .does not exist." This article describes a computer-aided software development system that does exist and is based on the proposed automation based paradigm. Although Balzer, Cheatham, and Green used the maintenance problem as the basis for asserting that the current paradigm has fundamental flaws, they could have used any portion of the software life cycle to make the same argument. For years the industry and end users alike have recognized that there is a software crisis. Symptoms of this crisis are visible everywhere: costs overrun; software does not meet user expectations; productivity is not greatly increasing; software specifications do not reflect the code and are allowed to become outdated; and the list goes on and on. Although the symptoms are visible, the cause of the software crisis has been less apparent. As a result, response to the crisis has been to treat the symptoms by educating programmers, attempting to enforce discipline, providing software tools, and creating new coding languages. All of these approaches do help to keep the software crisis under control, but they do not treat its cause. The cause of the software crisis is code. Under the current paradigm, the only formal specification language is code, and the only formal specification is the program. Although modern programming practices have defined techniques to make the program more readable, the fact remains that the program is far removed from the user. Programs show only detailed logic, and from that view, it is difficult even for programmers to glean an understanding of the performance characteristics ofthe software. As a communication vehicle, the program listing is analogous to attempting to describe a television set in terms of only schematic drawings. To the user, the program is an attempt to describe a picture using only words written in a foreign language, and the price of the discipline enforced by modern coding languages is that the language has become more foreign. Under the current …