Managing design trade-offs for a program understanding and transformation tool

Abstract

Software maintenance is costly. Part of the problem is that the repeated modification of a software system degrades its structure, thus making it difficult to understand and modify. Semantically rich techniques can help a software engineer understand or restore the structure of a system, but they typically use concurrency analysis or dependence analysis on pointers, which are difficult, at best, to implement efficiently. We believe that less expensive techniques can be used if the design and domain knowledge of the software engineer can be used in the analysis. Exploiting this knowledge may entail iteratively refining a query to find the right information, which requires a fast, flexible tool. We describe the design and use of a fast, programmable tool that can perform syntactically oriented text-processing tasks for use in program understanding and transformation. We take a “traditional” compiler approach to the problem to provide a tool with the flexibility and speed of UNIX tools such as AWK, but make structural adjustments to the tool's design to accommodate interactive analysis of an entire system. Our performance measurements suggest that this approach can produce results substantially faster than previous approaches.