configuration management

By reworking lean principles for the branching and merging arena, we're able to create automated builds and unit tests to increase effectiveness and improve quality in software configuration management. Individual developers and teams alike can benefit from this process-improving strategy.

One of the problems with Configuration Management (CM) and Project Management (PM) is that the tools and the data repositories for each are separate. As a result the processes are quite separate. The project manager takes a set of requirements and decomposes them into tasks, which are then prioritized, scheduled and assigned. The CM team creates configuration items and tries to tie the CIs back to the requirements so that they can be properly audited. When CM and PM work together they tend to enhance each others function, and eliminate potential overlap in the processes.

The promises of faster, better, and cheaper testing through automation are rarely realized. Most test automation scripts simply repeat the same test steps every time. Join Ben Simo as he shares his answers to some thought-provoking questions: What if your automated tests were easier to create and maintain? What if your test automation could go where no manual tester had gone before? What if your test automation could actually create new tests? Ben says model-based testing can. With model-based testing, testers describe the behavior of the application under test and let computers generate and execute the tests. Instead of writing test cases, the tester can focus more on the application's behavior. A simple test generator then creates and executes tests based on the application's modeled behavior. When an application changes, the behavioral model is updated rather than manually changing all the test cases impacted by the change.

Software that performs well is useless if it ultimately fails to meet user needs and requirements. Requirements errors are the number one cause of software project failures, yet many organizations continue to create requirements specifications that are unclear, ambiguous, and incomplete. What's the problem? All too often, requirements quality gets lost in translation between business people who think in words and software architects and engineers who prefer visual models. Joe Marasco discusses practical approaches for testing requirements to verify that they are as complete, accurate, and precise as possible-a process that requires new, collaborative approaches to requirements definition, communication, and validation.