物體導向軟體之複雜性量測之實驗研究

Abstract

物體導向程式規劃已經廣泛且逐漸被認識到可產生更能再利用之程式及減 少程式發展時間。複雜性普遍被用為一種估算發展程式之資源花費及預測 程式之可靠性和維修性。我們已發展一組以資訊流為基的物體導向程式之 複雜性量測。此論文中並提出一個實驗研究，設計來證實此一組量測。此 實驗被指導來計算用Ｃ＋＋語言所寫的方法，類別，類別階層和整個程式 之複雜性和發展成本。結果顯示某一程式對象用我們提出的量測所得出複 雜性與此對象發展時所花費的時間，產生的錯誤數目，修改的次數不僅相 關且敏感。對於類別階層和整個系統之複雜性測量似乎由其元件方法或類 別之複雜性加總較好。 Object-Oriented programming (approach) has been recognized prevalently and increasingly for its advantages of producing more reusable programs and reducing developing time. Complexity was commonly used as an attribute for assessing the resources spent and for predicting reliability and maintainability of a program. We have developed a set of complexity metrics, based on information flow, for object-oriented programs. This thesis presents an experimental study designed to validate the metrics. The experiment was conducted to calculate the complexity and development cost of methods, classes, class hierarchies, and programs in C++ language. The results show that the complexity of a program entity that is measured by the proposed metrics is correlated with and sensitive to the efforts, number of errors, and number of changes of the entity during programming development. The complexity measurement of class hierarchy and system seems better to be done by summing from the complexity of their component methods or classes.