This award is for continued research on the design and implementation of tools for interactive computer programming. Specifically, support is to continue the research being carried out by the Wisconsin Program-Integration Project, which is currently supported by the National Science Foundation under PYI grant DCR-8552602. The goal of the project is the development of a tool for semantics-based program integration. By program integration is meant the merging process that one has to go through when a program's source code diverges into multiple variants (supporting different features, different operating systems, or incorporating different bug-fixes, for example). Semantics-based program integration means that the tool must determine whether the changes that were made to the base program to create the variants have undesirable semantic interactions. (Becuase problems such as program equivalence are undecidable, an integration system must use techniques that are safe approximations to such undecidable problems.) The research explores ways to create a system that tests whether several different, but related, variants of a base program interfere, and, if there is no interference, automatically merges tham into a common version that combines their separate features.
