Using Program Adaptation Techniques for Ada Coverage Testing

Software development processes often employ program metrics tools to analyze software. Examples include static program analysis, code profiles, and coverage analysis. Automated tools can facilitate this study by collecting data and generating the desired measures. Unfortunately, the implementation of such tools has proven to be difficult and time consuming. In order to simplify the task of constructing analysis tools for Ada, we have examined a new parser generator system for use in a source to source transformation approach. Our goal is to allow developers to easily and compactly express what information they need from the input language in order to compute the desired statistics. One result of our research is AINT, an instrumenter and tester for Ada source code which was constructed using the new translation technology. AINT provides both statement and branch coverage testing of Ada programs, allowing users to evaluate the adequacy of their test data. This paper summarizes the new parsing system (called New Yacc) used to build AINT, then describes the implementation of our testing system.