re: How do you measure and discuss the less measurable things about testing code? VIEW POST


I mainly measure test coverage in terms of use-case coverage. Are we covering the primary app features, and their combination with other primary features.

For algorithmic modules I try to cover logical edge-cases but rely heavily on code review to ensure correctness.

I place no value on actual code coverage. It's a pointless number to chase and in no way relates to the use-case coverage. In terms of use-case coverage there is no actual upper limit to coverage, unless you have a trivial app. Orthogonal features have an infinite number of ways to be combined. It's just foolish to think test cases can somehow cover all of them.

I wait for complex edge cases to arrive and be reported by users before worrying about them. It's a game of priority. I write good code and hope it's generally correct. Realisitic priorities don't allow me to ensure it works 100%.

code of conduct - report abuse