Agile versus conventional development

Agile versus conventional development

There is a community of software engineers who practice what is called agile development. Agile methods emphasize the ability to quickly modify software and have been found to work well for small to medium-sized systems. The most well-known such method is called eXtreme Programming (XP). We will contrast agile methods with more conventional methods at several places in this book.

One way in which agile and conventional methods differ is in how they treat requirements and design. Agile practitioners gather requirements in very small increments, and design and implement each increment before gathering the next small requirements increment. They fully acknowledge that this may require the design to be changed to accommodate the new requirements, and use techniques called refactoring to make the necessary design changes. Conventional practitioners, on the other hand, prefer to develop a design that will be robust in the face of changing requirements. We will revisit all these ideas at various points in the book.

Quite often, for large systems, software engineers work on architectural design in conjunction with high-level requirements. This allows them to divide a system effectively into subsystems. Detailed requirements can then be developed for each subsystem. For smaller systems and lower-level subsystems though, it is conventional to develop the requirements before starting the design since otherwise the design may have to be re-done if requirements change.