[Haskell] Re: Functional dependencies and type inference (2)

[oleg at pobox.com]

Stefan Monnier wrote:
> > instance CpsForm t t where
>
> This can't be right, can it?

In general no: the CPS of a function certainly doesn't fit the above
pattern. So, if the source language has abstractions (the language in
the original message didn't), we have to add another instance for
CpsForm. Actually I started designing such a general typeful CPS of a
language with abstractions, applications (and perhaps even call/cc),
but then I ran out of time...