Representing and accessing extracted information

Source code repositories best support program, comprehension tasks when they can successfully answer the many questions a maintainer conceives. To provide the flexibility needed to answer these questions, the Jupiter repository system has been developed. Jupiter using Maia, a model based on annotations, indices and attributes, can store any of the syntactic, type and flow information extractable from a program. Jupiter's query language, Mercury, formed by merging an existing query language with Scheme, is used to access the repository and manipulate query solutions. Together these components provide a foundation on which to build systems for solving the queries that occur during program comprehension.