@hackage type-spine0.1.1

Until http://research.microsoft.com/en-us/people/dimitris/fc-kind-poly.pdf reaches the mainline, this is a surprisingly effective workaround. We support a limited number of kinds out-of-the-box, but it can be extended by the power-user. Also, quasiquotation makes the code rather legible. Given a finite set of kinds to support, generic type families can be defined that will work for an infinite number of types. It is very much a "bumping up" of the term-level spine view.

See the type-cereal package for a use case.