Building Maintainable Software, Java Edition

Appendix A. How SIG Measures Maintainability

SIG measures system maintainability based on eight metrics. Those eight metrics are discussed in Chapters 2 through 9. Those chapters include sidebars explaining how SIG rates source code properties relevant to those guidelines. These ratings are derived from the SIG/TÜViT¹ Evaluation Criteria for Trusted Product Maintainability. In this appendix, we provide you with additional background.

Together with TÜViT, SIG has determined eight properties of source code that can be measured automatically. See “Why These Ten Specific Guidelines?” for how these properties have been chosen.

To assess maintainability of a system, we measure these eight source code properties and summarize these measurements either in a single number (for instance, the percentage of code duplication) or a couple of numbers (for instance, the percentage of code in four categories of complexity, which we call a quality profile; see “Rating Maintainability”).

We then compare these numbers against a benchmark containing several hundreds of systems, using Table A-1 to determine the quality level on each property. So, if the measurement for a system is among the top 5% of all systems in the benchmark, the system is rated at 5 stars for this property. If it is among the next best 30%, it rates 4 stars, and so forth. This process of comparing quality profiles for each system property against the benchmark results in eight star ratings, one for each system property.

Table ...