Writeups of Experience with Code Coverage Tools

Created and summarized by Brian Marick.

Code coverage tools measure how test suites exercise programs. For example, which lines of code in the program were never executed? A line of code that has never been executed can hardly be said to have been well tested.

I wrote my first code coverage tool around 1982. I have since written three more. I have a fair amount of experience with coverage tools. I think these three writeups will give you a sense of what it's like to use code coverage. They are all PDF files.

• A Tutorial Introduction to GCT steps through a typical use of a coverage tool. It shows the mechanics. It assumes you can read C code.

• But how much does coverage cost? Experience with the Cost of Different Coverage Goals for Testing¹ shows that the most common types of coverage don't add much cost to thorough developer testing. Although this paper doesn't discuss "black box" testing of whole products through their external interface, I use coverage in much the same way in those situations: I design tests to find bugs, then use coverage to quickly look for test design mistakes.

• Coverage, when used, is often misused - by developers, independent testers, and managers. In How to Misuse Code Coverage², I describe common misuses. What's more, my explanation of how best to use coverage is more explicit than in the previous two writeups.

¹Experience was first published in the proceedings of the 1991 Pacific Northwest Software Quality Conference.

²Misuse was first published in the 1999 International Conference and Exposition on Testing Computer Software

Related Testing Craft Pages

• There will at some point be case studies in the index of techniques.

Be the first person to add a comment in the Wiki Forum at page CodeCoverage.
(The Forum is explained in its FrontPage.)

Summarized Discussion

In this spot, the author of this page will occasionally summarize the discussion in the Forum.