Theory and practice of middle-out programming to support program understanding

Theories of top-down and bottom-up program comprehension have existed for several years, but it has been recognised that understanding rarely happens in practice in such a well-ordered way. The paper describes recent work and results at Durham on what is termed middle-out programming. The objective is to avoid the problems of top-down and bottom-up approaches, by designing a very high level language specific to the application domain. Domain knowledge is captured in the design of this language, which retains a strong formal basis. This paper takes the view that software engineering will become strongly application domain based, and that knowledge representation of the domain will be a crucial factor in supporting program comprehension. An example of using this approach in the design of a large software system is presented