lox-haskell VS linear

If you feel like comparing implementations, you might want to take a peek at my implementation: https://github.com/Martinsos/lox-haskell -> I haven't finished the book unfortunately but I did all the work till chapter 9.5 -> each commit is marked with the chapter it implements, so you can easily follow the development that way. If you do make significant progress, do consider opening an issue on my repo to share your work, I would also love to do some comparison in order to learn more!

Very cool! More than a year ago I started going through Crafting Interpreters and implementing it in Haskell as I go, here it is in case you find it interesting: https://github.com/Martinsos/lox-haskell .

Linear.Quaternion.Quaterion from the linear package has a Num instance, and quaternion multiplication is noncommutative.

That's pretty much exactly the abstraction used in the 'linear' package's Affine class, in which we find points separated by vectors.

Linear algebra: hmatrix again with its own array type (and linear but this is restricted to low dimensions)

This is very niche, but something I've wanted to do for a while is to generate some cool physics example on the surface of a sphere with https://hackage.haskell.org/package/hamilton, and display it with https://reanimate.github.io/ (using https://hackage.haskell.org/package/linear for the projection)

Do check out the linear package. It has some complicated parts, but the types work out very nicely. Your Coords type is called V3 in that package, it has a huge list of instances for all kinds of different type classes.

I've come to love the linear approach, where instead of working with Naperian (Representable) functors, you work with Distributive functors.

See also the linear library, it contains types like data V3 a = V3 a a a with all of the useful instances.