After measuring 12,000 projects, a leading researcher reveals his findings.

Written by Steve Kilner

Maintenance and enhancement projects for legacy systems are notorious for being budget busters and deadline killers. This article gives a quick overview of how you can significantly improve the success rate of such projects. For more information, you can also watch these short videos: Using Complexity Analysis to Lower Project Costs and Hyper-Navigation of RPG Systems.

Categorizing Maintenance Projects

In his epic book, Estimating Software Costs, Capers Jones describes four types of maintenance projects. These should be familiar to experienced maintenance developers and managers.

Figure 1: Maintenance projects can be divided into four types. (Click image to enlarge.)

Struggling Productivity (And a Breeding Ground for Defects)

It should come as no surprise that productivity of Type 4 projects is lower than Type 1 projects, which is lower still than greenfield projects. The statistics that have been gathered in this book show productivity to be 5-10 times lower for Type 4 projects than for Type 1. Also critical for software maintenance work is the fact that more complex projects are notorious for injecting new defects into existing systems.

A Prescription for Maintenance Productivity and Fewer Defects

The clear solution that Capers Jones offers to these problems—as a result of measuring over 12,000 actual software projects—is that improvements are readily obtained through the use of tools—in particular, static analysis tools and complexity analysis tools.

Let's look at these in more detail.

Static Analysis Tools: Faster, More Accurate Impact Analysis and Fewer Defects

Questions maintenance programmers must ask include:

• What does this program do?
• What programs reference this file or field?
• Where do I need to change the code to produce the desired effect?
• What will happen if I change this statement?

Answering these types of questions can be very time-consuming and error-prone. Mentally tracking the paths of fields through subroutines, programs, and files is a tedious exercise often derailed by interruptions and human limitations of concentration and memory. Fortunately, much of this activity can be automated through the use of a top-tier static analysis tool. For a quick view of how a static analysis tool can facilitate program analysis, see this video: Hyper-Navigation of RPG Systems.

As has been documented by numerous studies, maintenance programmers spend approximately half of their time trying to understand existing code. Providing tools that augment their mental tasks can significantly improve their throughput and quality.

Complexity Analysis Tools: Better Planning, Estimating, and Budgeting

How often have previously unknown complexities revealed themselves in mid-project, thereby seriously damaging your timelines and estimates? This is a common problem in software maintenance.

These hidden complexities can be discovered earlier through better analysis, as described above, but they can also be factored into the plan in a more holistic way by understanding the complexity of the targeted code through use of complexity analysis tools.

Complexity analysis, sometimes referred to as software metrics, can measure many aspects of software. For planning purposes, a manager is helped by understanding the degree of complexity of the programs being modified.

First, a quick explanation of complexity. These are the most common factors used to measure complexity in most languages, including RPG:

• Quantity and depth of conditional logic
• Total and unique counts of variables and operations
• Amount of commenting available to the programmer
• Numbers of statements in programs and subroutines

It is one thing to add, change, or delete 10 lines of code in a 50-statement program, but it's quite another to do the same thing in a 5,000-statement program. By understanding complexity measurements of the programs and subroutines being changed, the project manager can factor risk premiums into the project. Risk premiums can be applied to these tasks:

• Completing program analysis
• Design activities
• Coding and unit testing
• Test planning for scoping and preparation
• QA and system testing execution

For a quick video demonstration of how to obtain complexity measurements for a project and factor them into a simple project spreadsheet, see Using Complexity Analysis to Lower Project Costs.

Overcoming the Challenge

To quote the man who has measured 12,000 software projects, Capers Jones, "Estimating the various forms of maintenance and enhancement work is among the most difficult problems facing software developers."

And he himself quotes 18^th century wit, Samuel Johnson, who upon seeing a dog walk on its hind legs, said, "It is not done well, but it is surprising that it is done at all."

as/400, os/400, iseries, system i, i5/os, ibm i, power systems, 6.1, 7.1, V7, V6R1

Steve Kilner is a leading expert in the field of static analysis, software complexity metrics, and program comprehension. He is a specialist in the maintenance and migration of legacy software systems for the IBM i and has written extensively on the subject. As the founder of and current CEO of vLegaci, he developed a combination static and dynamic analysis tool for RPG. He currently leverages his experience through his partnership with Databorough, makers of X-Analysis.