I looked into this a little bit and it looks like we have to special-case a bunch of things since an empty repository is qualitatively different from a nonempty repository (e.g., there are no branches) and some of the error conditions we're hitting are existing tests for other misconfigurations (e.g., you ignored every branch by accident or are trying to track branches which don't exist like "msater") so we need to add code to distinguish between the failure mdoes. We should definitely fix this at some point but the reward-per-unit-of-effort tradeoff isn't the best ever.