Software evolution in open source projects—a large-scale investigation

  • Authors:

  • Stefan Koch

  • Affiliations:

  • Institute for Information Business, Vienna University of Economics and BA, Augasse 2-6, Vienna A-1090, Austria

  • Venue:
  • Journal of Software Maintenance and Evolution: Research and Practice
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, the evolution of a large sample of open source software systems will be analysed. The evolution of commercial systems has been an issue that has long been a centre of research, thus a coherent theoretical framework of software evolution has been developed and empirically tested, most notably the laws of software evolution. In exploring the evolutionary behaviour of open source systems, these results can serve as a point of reference, allowing to assess if differences exist, or which aspects of open and collaborative development styles have an impact on evolutionary behaviour. The data collection method relying on a large software repository and the respective source code control systems is described, and an overview on the collected data on several thousand projects is given. The evolutionary behaviour is explored using both a linear and a quadratic model, with the quadratic model being shown as better suited. The most interesting fact is that while in the mean the growth rate is linear or decreasing over time according to the laws of software evolution, a significant percentage of projects is able to sustain super-linear growth. There is a positive relationship between the size of a project, the number of participants, and the inequality in the distribution of work within the development team with the presence of super-linear growth patterns. On the other hand, there is evidence for a group of projects of moderate size which shows decreasing growth rates, while small projects in general exhibit linear growth. Copyright © 2007 John Wiley & Sons, Ltd.