New metrics needed for new generation: lines of code, function points won't do at the dawn of the graphical, object era - computer-aided software engineering: includes related article on Windows test tools and a glossary on metrics for object oriented design - Buyers Guide

Software Magazine, May, 1992 by Jessica Keyes

For programmers who are not keen on moving to C to develop GUI implementation systems, more familiar languages are adding features of C development. For example, Micro Focus, located in Palo Alto, Calif., one of the beacons of PC Cobol, will soon make available a version of Cobol that permits Cobol programmers to have multiple open windows. One window may be for the text editor, another for compiling and another for debugging.

"So, even if a programmer keeps working in an old, familiar kind of language and methodology, and building familiar applications, he'll be doing it in an environment that is a lot more pleasant and productive to work in," said Yourdon.

Moving into these new paradigms of development presents the IS organization with a measurement quandary. What type of metrics should be utilized to effectively measure development? Is the old standard--lines of code--dead? Has the new kid on the block--function points--lost its utility? "The measurement field hasn't been changing very much at all to incorporate these new technologies," said Howard Rubin, a professor of computer science at Hunter College in New York City, and president of Howard Rubin Associates, Pound Ridge, N.Y.,

Although he stands behind the use of function points as a measure, Rubin admits that function points have limitations in dealing with more GUI-oriented systems. "Function points were created with a transaction notion of input and output. What do you do with the notion of color and motion? This is not reflected in the internal measure," he explained.

Today's systems, said Rubin, "very much stress the envelope of measurement." He added that no good, solid standards of measurement existed for hostcentric, terminal applications. "People are now trying to build on those very same things to deal with GUIs and object-oriented systems. There's a risk that a lot of technical size measures might start to fall apart."

Recognizing the inadequacies of current measurement practices for today's more complex systems, Rubin's team has developed a consumer reports scorecard structure." Rubin's scorecard includes 13 categories of measures, which can then be further refined down to 156 business-oriented measures.

Not all metrics are appropriate to all methodologies. For example, said Rubin, "People are trying to stretch function points to work with the object-oriented world. I don't believe it's stretchable. We might as well use something new. We can't always use the past for the future."

Research is underway at the Massachusetts Institute of Technology (MIT), Cambridge, Mass., to develop metrics for the object-oriented world. Dr. Chris Kemerer and Dr. Shyam Chidamber, professors at MIT's Sloan School of Management, recently co-authored a paper entitled, "Towards a Metrics Suite for Object-Oriented Design." Kemerer presented this paper at the October 1991 ACM Object-Oriented Programming, Systems, Languages and Applications (Oopsla) conference. In this paper, Kemerer asserts his position as perhaps the first person to talk about mearurement for object-oriented systems.