Metric based testability model for object oriented design (MTMOOD)

Raees Ahmad Khan, Khurram Mustafa

ACM SIGSOFT Software Engineering Notes · 2009 · 47 citations · 19 references

Concepts

TL;DR

Testability is a key factor for software quality, and predicting class testability early in the development cycle is increasingly emphasized. The paper reviews object‑oriented software testability and presents class‑level testability insights. It models the influence of design properties on testability by weighting their anticipated impact, building on a comprehensive review of class‑level testability. The authors propose the MTMOOD model and a suite of object‑oriented metrics to identify error‑prone design parts, show that early detection improves product quality and reduces testing effort, and validate the model empirically on industrial projects.

Abstract

This paper does an extensive review on testability of object oriented software, and put forth some relevant information about class-level testability. Testability has been identified as a key factor to software quality, and emphasis is being drawn to predict class testability early in the software development life cycle. A Metrics Based Model for Object Oriented Design Testability (MTMOOD) has been proposed. The relationship from design properties to testability is weighted in accordance with its anticipated influence and importance. A suit of adequate object-oriented metrics useful in determining testability of a system has been proposed, which may be used to locate parts of design that could be error prone. Identification of changes in theses parts early could significantly improve the quality of the final product and hence decrease the testing effort. The proposed model has been further empirically validated and contextual interpretation has been drawn using industrial software projects.

References

19