March 20, 2011, 5:12 a.m.
posted by freelancer
Design Approach and Artifact Choices
In the previous chapter, we looked at an XP-based approach to defining business requirements and working with the customer. In this chapter, we will drill down into some minimal architecture and design to help us get going with building Time Expression, using popular technologies such as Hibernate, the Spring Framework, the Eclipse SDK, and many other related tools such as Ant, JUnit, and more.
If you have come across the myth that XP programmers don't design or document, I hope this misconception will be cleared up by the end of this chapter, because it couldn't be further from the truth. Let me give you a preview of what I'm talking about.
Take a look at Figure, which shows some possible artifacts you can produce at the release or iteration level. Release-level artifacts are ones you produce prior to a new release; iteration-level artifacts are ones you produce prior to each iteration. These aren't all mandatory for every project, so we can pick and choose the ones we need. However, between Chapter 2, "The Sample Application: An Online Timesheet System," and this chapter, I have chosen to demonstrate as many of these as possible, and practical, for Time Expression. At the end of this chapter, I will show you another diagram that will tie together all the artifacts produced as a result of our efforts between the previous and this chapter (but don't cheat by looking now, because it is a detailed diagram and I don't want to overwhelm you at this point).
1. XP/AMDD style choices for artifacts to produce at release and iteration levels.
At the very least, what you will see in this chapter will give you one perspective. This process might or might not work for you. However, there must be some things good about these methodologies, because developers love them, and I have seen many successful projects as a result of these methods. Also, in our case, the artifacts we will produce in this chapter are essential to the rest of this book, and this process will help get us there.
As you can see from Figure, we have a few artifacts to produce in this chapter, so let's move forward. However, before we do, I want to provide two perspectives from real-world users of XP.
A project director working at a Fortune 50 company told me recently, "When we kick off an iteration, the first day of the iteration is usually spent reviewing stories and breaking them into tasks. The exercise of breaking them into tasks is truly a design session. What we ended up observing is that something like 20% of the developer's time, during an iteration, was spent in design. If you add all that time for all developers, across all iterations, it was a large numberwhich truly debunked the 'no design' comments."
To give you another perspective on the XP style of working, consider this statement from a senior architect at a well-established IT solutions company that has deployed more than a dozen successful projects using XP and AMDD techniques: "There is also another level of design that happens on an XP project which is at the daily level. Refactoring is a design activity. Although the iteration-kickoff design is an important step, it is the design work after the code is written that makes the difference between an OK design and a truly elegant one."
The difference with the XP approach, is that the architecture and design happens throughout the application's release cycle, not just up front. In other words, the application continues to evolve through the various iterations. The benefit of this approach is that the design is actually applicable to what you are building, not three to six months into development when the requirements could have changedsomething that is certainly possible in our fast-paced and ever-changing world today.