Published in Proceedings of the 11th IASTED International Conference on Software Engineering and Applications: Cambridge, MA, November 19, 2007, pages 7-12.
Software maintenance consumes around 70% of the software life cycle. Improving software maintainability could save software developers significant time and money. This paper examines whether the pattern of dependency injection significantly reduces dependencies of modules in a piece of software, therefore making the software more maintainable. This hypothesis is tested with 20 sets of open source projects from sourceforge.net, where each set contains one project that uses the pattern of dependency injection and one similar project that does not use the pattern. The extent of the dependency injection use in each project is measured by a new Number of DIs metric created specifically for this analysis. Maintainability is measured using coupling and cohesion metrics on each project, then performing statistical analysis on the acquired results. After completing the analysis, no correlation was evident between the use of dependency injection and coupling and cohesion numbers. However, a trend towards lower coupling numbers in projects with a dependency injection count of 10% or more was observed.
Published by Acta Press: Anaheim, CA.