CASE: a testbed for modeling, measurement and management

Reviewer: John E. Martin

A special series of six papers in this issue of Communications of the ACM addresses the general area of CASE as it relates to software quality, the software development process, and management attitudes and expectations. Norman and Forte (Introduction) As editors of this special CASE feature, Norman and Forte begin by discussing the heritage of the papers: the International Workshop on CASE (IWCASE), held in 1990. This short introductory piece then finishes with a brief abstract of each of the remaining papers. Forte and Norman (Self-assessment) As the title indicates, this paper presents an assessment of CASE. The assessment is a summary of opinions and perspectives from over 200 experts in software development technology from across the academic, supplier, and user communities. Key conclusions of this assessment include: Tighter integration among the various CASE tools is needed. Quality is the prime objective for pursuing CASE. Achievement of quality incentives has plateaued, however, because of the lack of a standard CASE methodology. A class of CASE support that has been overlooked is decision support tools for software portfolio analysis, buildbuy analysis, maintainrebuild analysis, and reusestart-from-scratch analysis. Shepard, Sibbald, and Wortley A number of formal software process models depict the relationships among the software development life cycle (SDLC) phases. In practice, many more models exist, because every software developer or maintainer uses a unique process. Because of this variety, these authors conclude that next-generation CASE environments need the ability to create a customized model of the software development process being used. This metamodel will represent activities performed completely by the computer, those performed completely by people, and those that involve human expert decisions aided by computerized tools. CASE tools used in conjunction with a software process model will help, encourage, or force users to follow all phases of the SDLC. Huff The author addresses the issue of providing a basis for developing a cost estimate for the adoption of CASE. Using and citing earlier works in this area, Huff describes a three-dimensional cost model that, within each phase of the SDLC ( x ), identifies cost items ( y ) and the cost drivers ( z ) that influence costs. Working through the model, he shows that the total cost of acquiring CASE may be five to eight times greater than the cost of the CASE tools alone. Maiden and Sutcliffe The reuse of specifications, according to the authors and the studies they cite, can help to overcome scoping difficulties encountered during the early stages of the SDLC. Additionally, in the hands of experienced engineers, reusable specifications can allow more use of prototyping in the analysis phases of projects. Tate, Verner, and Jeffrey Just as CADCAM has brought integrated design tools to the engineering of physical systems, CASE is bringing analogous tools to the engineering of the more abstract software systems. When integrated with metrics and a suitable software improvement model, CASE can increase the rate at which organizations improve their software engineering capabilities. Baxter Traditional software maintenance, in the absence of design information, has proven difficult. The archiving and recall of design information can serve to improve the maintenance process, however. The author contends that this feature alone may justify the investment in CASE. Conclusion Any organization looking to invest in CASE should first invest the time to read this nice suite of papers. Each work provides perspectives and information that can be used to develop a CASE investment strategy. The Huff paper, with its exploration of budget development, is particularly good. With the exception of the “Self-assessment,” all the papers include extensive references. All are well written and easy to read, with styles ranging from editorial to scholarly. The only perspective missing from this array is in the area of project team performance metrics. Much h as been written about estimating and performing tasks in a non-CASE environment. The authors might have considered this area as it relates to CASE. CASE vendors continue to claim fantastic productivity gains when using CASE tools, yet current research does not seem to support this claim. I would have liked to see a paper that addressed this productivity issue.