What are the factors affecting the handover process in open source development?

Abstract

BackgroundHandover is common in open source software (OSS) development, which could have a negative impact on software quality and progress. ObjectiveWe aim to identify factors associated with the handover process for future improvements. MethodWe first propose a metric, i.e. Active Days Coverage (abbr. ADC) together with an algorithm, i.e., Handover Duration Identification (abbr. HDI) to identify the handover processes in open software projects automatically. To evaluate our method, we selected two sample sets (i.e. sample set A and sample set B) from Github. With the sample set A, an automatic identification (to identify possible handover processes) together with an email inquiry (to identify actual handover processes) have been conducted. With the sample set B, we analyze fourteen potential factors impacting a handover process using the stepwise regression method. ResultsThe precision, recall and accuracy of our identification method reach 0.67, 0.73 and 0.65 respectively. The rate of correct identification of HDI algorithm is over 0.5 on average. Six factors were identified as the major factors impacting a handover process as well as seven combinations of these six factors were tested by stepwise regression method to explore possible correlation with the corresponding handover duration, among which five combinations show R2 greater than 0.4 with one reaches 0.493. ConclusionThis study implies that handover can be identified automatically. Developers usually follow a common handover process under various context. Moreover, although a significant correlation between the duration of a handover process and the combination of certain factors could be observed, there is no single factor that has a significant correlation with the duration of a handover process.