組込みソフトウェアの派生開発におけるソースコードメトリクスによる再利用性測定
ダウンロード数: 218回
年度 : 2012年  
発表場所 : SQiPシンポジウム
紹介文 :
修正なしのブラックボックス的な再利用の可能性を予測する枠組みを提案する論文です。再利用性を測定するために有効なソースコードメトリクスを抽出しただけでなく、さらにそれらのメトリクスが再利用性に与える影響の度合いまで明らかにしています。再利用される予定がないソフトウェア開発でも参考になります。
概要 :
Low understandability and changeability of embedded software could decrease the productivity and quality in later derivative developments due to long time of trying to understand and a broad range of modifications. The effects of static characteristics of source software on reuse have not been adequately discovered although it is preferable to evaluate ease of reuse without significant modifications in derivative developments. In this paper, we propose a framework for measuring reusability of C program source code by specifying static characteristics related to ease of black-box reuse without any modifications and analyzing relations to actual reuse rate in derivative developments. By applying the framework to 10 embedded software derivative development results in a construction equipment manufacturer, we obtained a set of reusability metrics mainly related to module couplings via external global variables; it is confirmed that these metrics could predict the actual reuse rate with high accuracy. As the actual reuse rate, we used the rate of the number of functions reused without any modification in total of the number of functions reused with or without any modification. This achievement is expected to be utilized for predicting future reuse rate and related efforts in derivative developments, and for supporting modification and/or refactoring of software under consideration of derivative development target. In the future, we will verify the generalizability of the framework and findings of the paper. Moreover we have a plan to consider the mechanism of proposing ways for improving reusability according to the measurement results.
↑