I had a dream. Instead of promoting unit testing against management and other developers I decided some months ago to get some experience in that field by my own and afterwards convince them with the facts, very similar to what Kevin Lawrence pointed out today. I planned to have a test for each and every new or changed feature.

Looking back at the last development cycle in my project I must say I didn't reach the goal. Initially I was very disciplined to at least try to setup a test. But the more the release date came closer the more I skipped writing a test for a feature.

Especially for complex database stuff (PL/SQL, complex views) it takes a lot of time to write good tests. Setting up data, checking the correct results and teardown the data is not a trivial task.

By the way, it is astonishing how little articles/experiences one can find about that topic on the net. Everyone praises doing unit testing but no one writes about how to do it.

Anyway, I keep on trying to get better in that field.