Measurement of change: stable and change-prone constructs in acommercial C++ system

摘要

Our previous studies of developers' ability to predict softwarechange revealed a great potential for improvement of the changemanagement process. What we consider most beneficial is characterizingand understanding software change by measuring it and identifying whatkind of changes take place and how frequent they are. With suchknowledge, it is possible to build change models that help developersmake better predictions regarding future requirements. An analysis ispresented of a commercial object-oriented system-the PMR (PerformanceManagement traffic Recording) system of Ericsson Radio Systems-that waschanged significantly due to the implementation of a set of newrequirements. Measures are used in order to identify stable andchange-prone constructs of the system. The results from the analysis arecomplemented with results from interviews with developers about whatchanges most frequently in a C++ system