Systematic Study of COBOL Programming.
FINAL REPT. 19 Sep 79-8 Mar 81,
PURDUE UNIV LAFAYETTE IN DEPT OF COMPUTER SCIENCES
Pagination or Media Count:
This report studies the applicability of the Theory of Software Science to COBOL Programs. A COBOL analyzer was written to produce the most important Software Science metrics. The analyzer was used to study several COBOL Databases. The analyzer has been distributed to AIRMICS and to several other organizations including Ohio State University, Jet Propulsion Labs, CINCOM INC., and the University of Trondheim, Norway. Using the analyzer we investigated the software Science Effort Estimator and compared this with estimator based on lines of code and on cyclometric complexity. Our experiments show that the software science effort estimator works reasonably well only if a program is properly modularized and if inter-module factors are included. Published reports on our findings are available on request. We also studied the language level hypothesis of software science in a COBOL environment. While the study shows that the mean language level for COBOL falls between those for FORTRAN and PLI, the study also shows that language level is heavily dependent on the size of the program and that it fluctuates too much to be considered constant as the Halstead theory postulates. Author
- Computer Programming and Software