How to model mixins / multiple interfaces in Haskell?

Question

I came across this question on modeling inheritance in Haskell and it reminded me that I have a little more complicated version of the same problem. I\'ll adopt the example from

Answer 1

Perhaps it's just that this example with colours isn't particularly good, but it seems to me that you shouldn't ever truely need this and it wouldn't actually be good if it worked.

Physical is indeed perfectly natural the way you propose it: a Monster, Camera etc. doesn't have a position by itself, rather position is what you get by combining such a object with some space to live in.

But Coloured is different, for colour is a property of the thing itself and will probably have quite different meaning for a monster compared to a camera, so unlike Physical a type class would actually seem reasonable here. If at all – perhaps it would actually be better to simply use monomorphic functions for dealing with the various kinds of colour-ness manually.

Of course, you might be tempted think of it this way: things themselves aren't coloured, but they wear a skin that has colour. I don't think this should be the only way to have colour, but... fair enough, we can obviously provide such a "skin" so uncoloured objects become colourful too:

data ClSkin a = ClSkind { clSkinColour :: Colour
                        , clSkinned :: a         }
instance Coloured (Clsskin a) where
  colour = clSkinColour

Now you say it shouldn't matter if you use Physical (ClSkin a) or ClSkin (Physical a). I say it does matter. Again, Physical is sort-of a combination between an object and the entire space it lives in. Surely, you don't want to colourise that entire space! So really, Physical (ClSkin a) is the only meaningful variant. Or, alternatively, you might say colour is something that only makes sense for objects in a physical space. Well, then you'd just make the colour an extra field of that data!

data Physical a = Physical a Vec3 Vec3 (Maybe Colour)