Heuristic Strategies for Recommendation of Exception Handling Code

Abstract

Software developers have difficulties in implementing exception handling code in their systems. In particular, they fail in implementing appropriate handling actions. Not surprisingly, recurring failures and performance issues are often related to the poor exception handling actions implemented in real software systems. In this paper we propose, implement and evaluate a set of three heuristic strategies used to recommend exception handling code in programming tasks. Given a method in which the developer wants to handle an exception, the heuristics recommend a list of code fragments implementing exception handling. The heuristics' goal is to accurately find code fragments implementing exception handling and recommend these fragments according to the context of developers' implementation task. Hence, the proposed heuristics may assist developers in the process of discovering exception handling actions relevant to their context by providing concrete examples. We believe that a recommender system based on the proposed heuristics may be used in the future as a tool for aiding exception handling implementation.