I’m testing web service and all my tests are automated, each have it’s setup (crate data), testing (do actions with data) and clean-up (delete data) steps, so that old data do not affect next test run. Quite often when test fails I investigate the reason and forget to clean up, so I have a special cleanup all utility class. However the bug I discovered because I forgot to clean up and it caused null pointer exception, because my test tried to do rename to the name that already existed.
I could have adjusted my test suite to add this case, but I didn’t because I think 1) it’s not important very test case, though I admit I missed this corner case when first designed my tests, but - 2) I tested it once during bug retesting and I think risk of breaking this later is close to none.
So what’s subconscious or by chance?
People keep telling me writing down manual tests are great idea because one could review them and suggest how to improve or what’s missing. I keep telling that whatever I write down it does not describe all the tests and certainly don’t describe all the assertions I’m doing during manual tests. I’ve blogged about it long before. Today I realized that even with automated tests it’s true – I do more testing that I test suite claim to cover, just because I sometimes make mistakes and do some that i did not expect myself to do..
Add a Comment