Mason Protter

I'll leave this open as a somewhat far-off speculative potential feature.

Hmm, this is actually a little strange. None of the unit systems listed in these wikipedia pages are what most practitioners I know would call natural units and instead are...

Interestingly, the wikipedia page even says > In this case, the reinsertion of the correct powers of e, c, etc., can be uniquely determined.[1][2] but this is just untrue for...

No, the choice of `4πG` is not what I'm talking about. If you set **any** multiple of `G` to `1`, then you collapse the unit system and have no unit...

This doesn't work for multiple traits though. I.e. what's desired is that `f` would dispatch on whether or not `x` has the trait `Add1` or `Add2`. ```julia using CanonicalTraits @trait...

I think there's no real reason to put the return type in trait parameters is there? You could just do ```julia julia> @trait Addable{L, R} begin (+) :: [L, R]...

It doesn't seem to work for me. Perhaps I'm doing something wrong? ```julia julia> using CanonicalTraits julia> inferAddable(::Type{Int}, ::Type{Int}) = Int inferAddable (generic function with 1 method) julia> @trait Addable{L,...

Ah, I see! Thanks for explaining.

The problem here is basically that base makes these functions type unstable because it's known that the overhead of working with regular heavy arrays makes that instability okay to work...

I might just remove the `benchmark` script. Sounds useful to have, but I don't know how to maintain it.