Encyclopedia of Computer Science and Technology

software engineering 443Padgham, Lin, and Michael Winikoff. Developing Intelligent AgentSystems: A Practical Guide. New York: Wiley, 2004.Software Agents Group (MIT Media Lab). Available online. URL:http://agents.media.mit.edu/. Accessed August 21, 2007.software engineeringBy the late 1960s, large computer programs (such as theoperating systems for mainframe computers) consistedof thousands of lines of computer code. In what becameknown as “the software crisis,” managers of software developmentwere facing great uncertainty about both programdevelopment schedules and the reliability of the resultingprograms.Programming had started out in the 1940s as an offshootof mathematics, just as the building of computers wasan offshoot of electrical or electronic engineering. Increasingly,however, programmers were searching for a new professionalidentity. What paradigm was truly appropriate?Should programmers strive to be more like mathematicians,seeking to rigorously prove the correctness of their programs?On the other hand, many programmers thought oftheir work as a craft, performed using individual experienceand intuition, and not easily subject to standardization.Between the two poles of mathematics (or science) andcraft came another possibility: engineering.The concept of software engineering proved to be attractive.Mathematics (and science in general) are usually carriedon without being immediately and directly applied tocreating a particular device or process. Outside of researchprograms, however, computer applications were written toperform real-world tasks (such as flight control) that havereal-world consequences. Thus, although the notation ofa computer program resembles that of mathematics, theoperation of a program more nearly resembles that of complexmechanical systems created by engineers. By attachingthe label of engineering to what programmers do, advocatesof software engineering hoped to develop a body ofpractices and standards comparable in some way to thoseused in engineering. Some critics, however, believe thatthis paradigm is inappropriate, either because they believeone should strive for the greater rigor of science or out of apreference for individual craft over standardization.Programming PracticesOne of the most pervasive contributions to software engineeringhas been in computer language design and codingpractices. At about the same time that the concept of softwareengineering was being promulgated, computer scientistswere advocating better facilities for defining and structuringprograms (see structured programming). These includedwell-defined control structures (see branching statementsand loop), use of built-in and user-defined kinds of data (seedata types), and the breaking of programs into more manageablemodules (see procedures and functions).The next paradigm came in the late 1970s and hadtaken hold by the late 1980s (see object-oriented programming).The ability to “hide” details of function withinobjects that mirrored those in the real world provided a furtherway to make complex programs easier to understandand maintain. The growing use of well-tested collectionsof procedures or objects (see library, program) has beenessential for keeping up with the growing complexity ofapplication programs.Software engineers are also concerned with developingtools that will better manage the programming process andhelp ensure that standards are being followed (see programmingenvironment). The use of CASE (Computer-Aided Software Engineering) tools such as sophisticatedprogram editors, documentation generators, class diagrammers,and version control systems has also steadilyincreased. Today many of these tools are available even onmodest desktop computing environments (see case).The Program Development ProcessPerhaps the most important task for software engineeringhas been seeking to define and improve the process bywhich programs are developed. In general, the overall stepsin developing a program are:The Waterfall, or Cascade, model sees software development as a more linear process going through the requirements, design, implementation,integration and testing, and maintenance phases. The results of each phase cascade down into the next.