When you encounter a bug, do you first write a failing test before fixing it? You should. But the vast majority of us sometimes do not. External pressure (read Customer, Teammates or Management) moves us in the opposite direction. Everything we’ve learnt about unit testing and good software practices go out the window, and the sheer pressure forces us to fix the issue as fast as possible! We focus more on getting the job done* and deploying than worrying about adding another test to our suite.
Adding that test is not about increasing our code coverage or patting ourselves on the back, with an optional tweet, that we are good developers. It’s about putting regression tests in place.
Of course, we could always just apply the fix, and then later, calmly add a test. However, there’s always another urgent bug fix, feature, or customer breathing down our necks, so things just get putt off.
It’s hard to not be tempted to just fix something then and there. Fight the temptation though if you can. Write that failing test first!
[*Implicitly changing the definition of Done on the go]
08-30-2010 2:24 PM