Application Lifecycle Management

We seem to be under a curse in our profession. Although not cast by a witch or a wizard, the curse affects us just the same. It is the curse of "either/or"-the curse that we must choose either "this" or "that" but we cannot choose parts of both. Nowhere is this more evident than in today's struggle between the adherents of the traditional "plan-driven" and newer "Agile" approaches to software development. What most overlook is that both groups want to achieve exactly the same goal: quality software that meets customer needs within the constraints of time, budget, staff, and technology. They differ only on the strategies to achieve this goal. For example, both groups agree that system requirements must be understood; their differences lie in questions of "how much of what to do and when to do it." Lee Copeland offers insights and suggestions on the methods and approaches that will be most valued on your project-control vs.

The Agile development movement has started to transform the software landscape. Since February 2001 when the Agile Manifesto was published, Agile development has gone past the early adopter phase and now is regularly in use by such mainstream organizations as CapitalOne, the Federal Reserve Bank, Microsoft, Sun Microsystems, and major development departments in other organizations. For Agile practices to take hold-and more importantly to be sustained-all of the dysfunctional behaviors that organizations have acquired over the past twenty years or more must be discarded-and that is not easy or fun. However, the overwhelming benefits and value of Agile development are there, and for organizations in which software is their lifeblood, the effort will be made.

Often times, deciding what code to refactor (rewrite) is based upon the code's smell, a subjective determination by developers. Learn about and see examples of static analysis techniques such as cyclomatic complexity, depth of inheritance, and fan-in/fan-out to augment developers' instinct with real data. For example, cyclomatic complexity is adept at spotting methods containing too much conditional logic. Deep hierarchy trees, problematic for testing, can be broken out into separate objects. Fan-in/fan-out analysis is effective at pinpointing brittle code that you should refactor. Armed with new methods to objectively spot smelly code and replace it with proven patterns, your code base will become more stable and maintainable.

The concept of software factories is becoming a hot topic in software engineering circles. So, how can the factory model fit with Agile development practices? Damon Carr makes the case that Agile development is a stepping stone-not an alternative-to software factories. This is not the dreary vision of mindless workers in a factory. Instead, think of highly skilled individuals working with multi-million dollar machinery to develop systems. Even if you are not considering the factory model, Damon offers new practices that can reduce overall Agile development costs by as much as 40 percent. These include explicit refactoring to design patterns in your iterations, quantitative risk management, metrics for understanding the health of your project, and a new approach to team structure.