Some tasks get defined, but never launched, because their task-dependencies are never met. As a result, the script is not run, which means it never has a chance to throw a failure, much less a success.
Our pipeline should try to track the # of unlaunched-tasks and offer up a summary of how much of the workflow is "unrealized" given a failure.