bzinberg

LMK if this looks like a good approach. If yes, then I'll add unit tests.

> If we'd like to deal with the problem correctly, I think it's better to add types for `UnitQuaternion`, `MRP`, (and etc.) and the types should not be subtype of...

I'm totally on board with implementing efficient conversions where possible, such as `UnitQuaternion(::RotX)` as you showed above. To me, that is independent of the inheritance question. Inheritance decisions among "public"...

Also, while I agree that the reasoning you give here is self-consistent: > I think the `==` operator here is for "equals as a matrix" because `UnitQuaternion

Thanks for this input @c42f, all of this sounds great. > if the perceived problems with `UnitQuaternion` could be solved with some documentation and a few careful renamings/deprecations that would...

Yeah, I figured as much -- thanks for shedding some light on this @fplk. Given that there are no known low-hanging fruit, and the current situation is tolerable, I think...

> * In `log-categorical`, adjust the weights in log space to prevent underflow. Probably a decent way to do this is by subtracting `(max scores)` from each entry of `scores`...

Looking into it. I think I may have been mistaken for Part 1, there's no path problem because `test/` is in the project's classpath. If that's correct, then Part 2...

The behavior I'm seeing seems to contradict the [Clojure docs](https://clojuredocs.org/clojure.core/require): > `'require` loads a lib by loading its root resource. The root resource path > is derived from the lib...

Ah, I think I'm onto something. The `test/` directory is in my classpath as reported by `lein classpath`: ```sh lein classpath | tr ':' '\n' # If there are a...