Jonathan Cast wrote:
So there is a program (or, rather, type) you can write with newtype that can't be written with data:
newtype T = T T
That compiles, and anything of type T is ⊥. But it breaks my mental model of what the compiler does for newtypes. I always think of them as differently typed versions that share the same underlying "data" declaration and representation; and then the compiler erases the type information. So let me think about this one. Looking at the Haskell 98 Report http://www.haskell.org/onlinereport/decls.html#sect4.2.3 "A declaration of the form newtype cx => T u1 ... uk = N t introduces a new type whose representation is the same as an existing type. The type (T u1 ... uk) renames the datatype t. It differs from a type synonym in that it creates a distinct type that must be explicitly coerced to or from the original type" What I don't see is anything that discusses what "newtype T = T T" could mean. Is there any difference in how GHC treats "newtype T = T T" versus "data T"? -- Chris