The Effect of Method Origin on Method Invocation Fault Frequency

Recent studies in analyzing bug-fix patterns recorded in software repositories show that method invocation(MI) fixes are very common. Our belief is that these bug fixes could be analyzed further to provide detailed information that allows software engineers to avoid these bugs. For example, developers may have a higher awareness of their intentions in choosing which method to invoke. Unfortunately, previous studies could not provide such information. In this paper, we propose atechnique to accurately identify the precise MI bug-fix changes. We have applied our technique to empirically investigate the origins of MI bug fixes in six open source Java software. We have classified these MI bug-fix origins (e.g. from standard Java library, from third-party library, or in standard code-base). Amajor finding is that MI faults involving Java library methods were under-represented when compared to their usage frequency.