Post

Periodic reminder: no matter how useful you think tests are, they are 16% more useful than that (a number I just made up). I wrote a feature, shipped it w/o tests, later added a bug but fixed the bug without adding a test. Adding a test later made clear a further improvement. Write more tests!

Using a tool like SimpleCov can make it really easy to determine code which has no testing. Now write tests which exercise that code, especially focusing on code which changes a lot. Change == bugs. github.com/simplecov-ru...

Are you using something to track coverage over time? I’ve tried a couple of the popular ones and either they are flaky or simplecov is flaky to the point of making it not valuable. Or, perhaps I have a skill issue 🫠

I only care about now. Haven’t used anything to track historically. I usually just check it when I’m changing an area and want to judge risk.