Software comprehension based on database relational algebra

Abstract

Execution trace file analysis facilitates software comprehension which is a fundamental part of the software maintenance process. However, the scalability of large execution trace files limits understanding programmes and increases the complexity of the maintenance process. A programme interior representation which reflects the actual data structure is constructed from reduced execution traces. These reduced execution traces are extracted from a profiled run or logs of a programme. As a result, the generated output of these representations can then be easily analysed and visualised. This paper presents an approach to analyse execution trace files using relational database algebra methods to better comprehend and maintain softwares. Furthermore, this work can be helpful in evaluating the software's security using the logs trace files. The proposed approach was tested on real open source software. The results indicate that the proposed approach simplifies software comprehension, and may reduce the execution trace file size for up to 70% of its original size.