An Empirical Study on the Usage of SQL Execution Traces for Program Comprehension

摘要

Several studies have investigated dynamic analysis in the context of software maintenance and evolution, and most of them confirmed the positive impact of such analysis on program comprehension tasks. In this paper, we focus on the understanding of the database access behavior of a program, which has become an important (yet largely ignored) aspect of program comprehension. We empirically assess how developers/students are (not) able to understand interactions between the database and the application program. To this end, we used DAViS, a tool for Dynamic Analysis and Visualization of SQL execution traces. We present a controlled experiment that quantitatively evaluates to what extent DAViS can influence program comprehension in terms of duration and correctness of the tasks. The results of the study indicate that DAViS does reduce the response time and increases the correctness (with a large effect size), which means that we found a strong indicator that the chosen approach is truly able to help developers.