Today, I managed to put in some time refactoring the punq fork. It's not perfect yet—there are still some type issues—but, um, huh. One other issue that seems like it shouldn't have any connection to what I just did. And, I've just discovered that the HTML report out of mypy is apparently totally orthogonal to its error reporting. That's... not what I expected, and I hope I can change that. Aha, looks like I want the --junit-xml option. I'll keep the other HTML report, as well.
Anyway, I'll test that out while this publishes. I'm pushing things late already, I should get to bed as soon as I can.