The evolution of technical debt in the apache ecosystem

Abstract

Software systems must evolve over time or become increasingly irrelevant says one of Lehman’s laws of software evolution. Many studies have been presented in the literature that investigate the evolution of software systems but few have focused on the evolution of technical debt. In this paper we study sixty-six Java open-source software projects from the Apache ecosystem focusing on the evolution of technical debt. We analyze the evolution of these systems over the past five years at the temporal granularity level of weekly snapshots. We calculate the trends of the technical debt time series but we also investigate the lower-level constituent components of this technical debt. We aggregate some of the information to the ecosystem level. Our findings show that the technical debt together with source code metrics increase for the majority of the systems. However, technical debt normalized to the size of the system actually decreases over time in the majority of the systems under investigation. Furthermore, we discover that some of the most frequent and time-consuming types of technical debt are related to improper exception handling and code duplication.