Measurement of Object-Oriented Software Development Projects

Executive Summary This technical report summarizes the state of the practice for measurement of object-oriented (OO) software development projects. This report extends work previously performed by the Software Productivity Consortium and National Research Council of Canada. It is one of a set of reports on measurement topics sponsored by the Department of Defense. These reports are intended to help those software intensive projects that are employing new technologies and methods to establish effective measurement systems for project management purposes. The main body of this report discusses the common issues encountered on object-oriented projects, provides recommended management measures, and explains how object-oriented measures and analyses can be integrated into the estimation and performance analysis activities described in Practical Software Measurement. Appendices provide surveys of the literature discussing a wide range of measures that potentially are useful for estimating software quality, project resources, and reuse. The final appendix identifies some of the tools available to support measurement. The general process for measuring object-oriented software development is the same as that for measurement of any other type of software. However, the specific size and quality measures, algorithms, and normative values of parameters must be adjusted to reflect the nature of the software and the development method applied. As many different object-oriented development methods are in use today, the availability of desired measures will vary from project to project. The planning, estimating, and tracking approach described here is based on that presented in Practical Software Measurement (PSM) [50]. PSM defines a generic process for selecting software measures and using the resulting information to manage projects. This report is based on a broad survey of the literature that which assesses the state-of-the-art and practice in object-oriented measurement and modeling, and maps the information collected onto the PSM framework. Estimation typically produces projections of the software size, effort, schedule, and quality required to complete the project. These estimates form the basis for initial project plans and subsequent replans. Generally, the same factors that get estimated need to be tracked in order to manage a project to a successful conclusion. Changes or mistakes in the assumptions driving the plans must be recognized as early as possible so that corrective action can be taken. Seven different approaches to estimation of object-oriented software projects are summarized in the appendices. Not all of these approaches cover all of the elements that need to be estimated for project planning. These approaches …