The period 2002-2004 was one of enormous progress in figuring out how testing fits in on agile projects. Test-driven design is more about designing and writing the code than about finding bugs. New testing tools such as xUnit and FIT came out and received a lot of use by early adopters. The hopeful notion that customers would write acceptance tests to find bugs was expanded, challenged, and deepened. With all that progress, it's hard to be dissatisfied with these methods in agile projects. But past ways of thinking are holding us back. To make further progress, we have to split our notion of testing into two parts: the task of after-the-fact product critique, and a role that has nothing at all to do with bugs and, really, little to do with the word "testing." Brian Marick, a founding member of the Agile Alliance, explains what that role presents and some ideas on how to fill it.

Design and architecture decisions made early in the project have a profound influence on the testability of an application. Although testing is a necessary and integral part of application development, architecture and design considerations rarely include the impacts of development design decisions on testability. In addition, build vs. buy, third party controls, open source vs. proprietary, and other similar questions can affect greatly the ability of an organization to carry out automated functional and performance testing-both positively and negatively. If the software or service is delivered to a separate set of end-users who then need to perform testing activities, the problems compound. Join Jay Weiser to find out about the important design and architecture decisions that will ensure more efficient and effective testability of your applications.

Using a challenging client engagement as a case study, Rex Black shows you how he and a team of test engineers created an integrated, automated unit, component, and integration testing harness, and a lightweight process for using it. The test harness supported both static and dynamic testing of a product that ran on multiple platforms. The test process allowed system development teams spread across three continents to test their own units before checking them into the code repository, while the capture of the tests provided automated integration testing and component regression going forward. He'll also explain the tools available to build such a testing harness and why his team chose the ones they did.

• Examine the benefits-and challenges-of implementing an integrated, automated component and integration testing process in a Java/EJB development environment

Testing an application's robustness and tolerance for failures in its natural environment can be difficult or impossible. Developers and testers buy tool suites to simulate load, write programs that fill memory, and create large files on disk, all to determine the behavior of their application under test in a hostile and unpredictable environment. Herbert Thompson describes and demonstrates new, cutting edge methods for simulating stress that are more efficient and reliable than current industry practices. Using Windows Media Player and Winamp as examples, he demonstrates how new methods of fault injection can be used to simulate stress on Windows applications.

• Runtime fault injection as a testing and assessment tool
• Cutting edge stress-testing techniques
• An in-depth case study on runtime fault injection