Measuring Commercial Program SuiteComplexity
David Leigh, E. Halliday · WIT transactions on information and communication technologies · 1970
COBOL is a very widely-used computer language, even in the 1990s. Its age is reflected in the ease with which poor initial programming practices are possible, even though the language does allow correct structuring. This may be contrasted with those languages of more modern derivation, where encapsulation, abstraction and data-hiding capabilities encourage correct approached. A large burden of maintenance also comes with older and less welldocumented program suites. As more and more modifications are made, the spaghetti nature of its code can become emphasised. To avoid this difficulty, monitoring of the complexity of the code can give early warning of troubles to come. Complexity measurement has generally been undertaken in languages such as FORTRAN, as being more readily available, and more in their approach. A similar approach to a non- scientific language has not, so far as is known, been attempted. The apparently free-form nature of COBOL can, however, be analysed efficiently, and the measurements applied throughout suites of programs. The structure of programs written in ANSI Standard COBOL (1985) 2 is described. The measurement of complexity for individual programs written in COBOL is discussed. The alternatives available for appropriate evaluation of complexity are investigated. The extension of complexity measurements to a suite of associated programs is described. The requirements of meaningful results are evaluated, with the practical and commercial uses to which they may be put. The application of these metrics to an existing program suite are described, with the effects arising from their use. The conclusions relating to their effectiveness, and further refinements to be made, are given. Transactions on Information and Communications Technologies vol 8, © 1994 WIT Press, www.witpress.com, ISSN 1743-3517