Derived functor context has type variable mismatch about derive

[ndmitchell]

Any reason for not doing:

{-# LANGUAGE DeriveFunctor #-}
module Test where
newtype WrappedFunctor f a = WrapFunctor (f a)
    deriving Functor

The deriving functor was added to Derive as a prototype for the GHC version. It's pretty complex, and testing suggests the GHC one got some improvements after that.

[RyanGlScott]

I'm not planning on using this in production code. I was just curious to see if derive was capable of handling this case, since I'm working on a similar Template Haskell-based project for deriving Bifunctor instances, so I wanted to take some inspiration from derive.

Am I correct in assuming that if derive sees a type variable of kind * -> * that it makes it an instance of Functor? Or does it do something more complex?

[ndmitchell]

I've no real idea. The code is at https://github.com/ndmitchell/derive/blob/master/Data/Derive/Functor.hs and https://github.com/ndmitchell/derive/blob/master/Data/Derive/Internal/Traversal.hs. I didn't write it, but you're free to reuse what you can. I suspect it's not looking at the kind info at all, since I don't think the version of TH that it was written provided that.

Should you wish to submit patches for your Bifunctor derivation to Derive and ship your derivation that way, I'd be happy to accept them.

Given that there's probably no great reason to use the Functor instance, and fixing it is probably not trivial, I'll close this ticket and just accept it as an expected limitation.