Maybe there's some kind of sampling bias, because the "obviously correct" code is usually easy to cover, so if I haven't covered a function completely, it's probably doing something horrible. Even if that's the case, I, um, don't care.
Just a little bit at a time, and it adds up.
"What if humans are the real monsters?" (Narrator: "Actually, monsters were the real monsters.")
It looks really complicated, but actually it can be very simple, unless I'm badly mistaken and it has to be extremely complicated.
Maybe I should re-evaluate the names before I work on the coverage? I don't know...
The bad names are still bad.
I think this went pretty well for this time of year. I will not check to confirm.
On the road again again.
It'd be cool if this were obvious a priori, but a posteriori is better than nothing.
It's really gratifying that there weren't problems where there weren't supposed to be problems.