[Haskell-cafe] haddock changes between 2.10 and 2.12?

Was there some significant change/bug introduced to haddock made between 2.10 and 2.12? When I look at the haddocks for kan-extensions 3.1: Data.Functor.Yonedahttp://hackage.haskell.org/packages/archive/kan-extensions/3.1/doc/html/Data... which purports to have been built by haddock 2.12 it shows all sorts of spurious superclass constraints on almost every instance that should have simplified out, but going back to kan-extensions-2.7: Data.Functor.Yonedahttp://hackage.haskell.org/packages/archive/kan-extensions/2.7/doc/html/Data... none of this extra noise is present -- and the code in question hasn't been changed in the interim, so I'm pretty sure its not on my end. -Edward