The example for GHC Generics is kinda broken
The popular GHC.Generics serialization example at http://www.haskell.org/ghc/docs/latest/html/users_guide/generic-programming.html\#idp25226064 illustrates how to serialize a sum datatype (a :+: b), with a 0 bit representing a and a 1 bit representing b.
However, consider a :+: b :+: c.
If the compiler treats this as (a :+: b) :+: c, then a is 00, b is 01, c is 1.
If it's a :+: (b :+: c), then a is 0, b is 10, c is 11.
The compiler's decision, even though (as I understand it) consistent for any given compile, could differ later.
The manual, at http://hackage.haskell.org/package/generic-deriving-1.6.3/docs/Generics-Deriving-Base.html\#g:9 states not to depend on the ordering, which the example does.
I think this is a valid concern, and the manual should be at least updated to discuss this problem.