Abstract

Query-based code browsers can be customized to render many different types of views on-demand. In contrast to single-purpose browsers offered by modern IDEs such as Eclipse (e.g. Call Hierarchy), they release developers from mentally inducting the integral information from several views generated for specific purposes. However, customizing query-based tools, and communicating with their interfaces demands database expertise and understanding of the query language syntax. This puts a burden on maintainers by forcing them to encode their required views using complex queries. In this dissertation we investigate a query engine designed for software structural analysis which (1) provides a visual query interface over the high-level textual query language to eliminate the need for understanding the query language syntax, (2) incorporates the knowledge of programming language constructs into the factbase, query language, and the views, and (3) integrates the query-based and specific-purpose views already provided through the IDE. We are confident that this approach will be beneficial to maintainers during comprehension by allowing to abstract source code to high-level views and to speed up the rummage of the source code.