In the test - encoding - reconstruction, it will feel a bit incredible when the preceding idea starts in contact with the front idea: If you do a test that does not have or will be wrong, is it a step forward?
In fact, see how you see this problem. Aiming at a failed test, a view is wrong; but another more positive view is to see the test and code as a system, a failure test shows at least a positive place: our system can Tell us when there is no such error / there is such an error, is this not a major progress, isn't there ancient training, "I know the shame,"? So let us cheer when testing an error, our system will enter a step forward, test again and again, a simple, clear, can measure the goal: modify the code, let the test pass .