Yesterday we presented our paper “Inferring Performance Bug Patterns from Developer Commits” at ISSRE 2019 in Berlin.
“Performance bugs” denote unnecessarily slow running code and have recently been the focus of many research articles in the software engineering community. In our paper we report on a manual investigation and classification of 733 performance bug inducing developer commits from 13 popular open source projects written in C or C++. In total, we extract 7 common performance bug patterns from these commits and analyze their differences in terms of bug complexity by looking at how long it takes before bugs are found and fixed, the seniority of the fixing developer, and how many lines the bug fixing commit comprises.
We hope that the results from our study will benefit those who are working on performance bug detection and localization techniques. An author copy of our paper can be found here.
The data we based our study on is publicly available on the project’s gitlab site. We appreciate any comments on this work and are happy to support further usage and enhancements of this data set.