[Haskell-cafe] Haskell type system and the lambda cube

[wren ng thornton]

voigt at tcs.inf.tu-dresden.de wrote:
> > 2009/5/24 Petr Pudlak <deb at pudlak.name>:
> > If all Haskell had would be HM, it would be System F.
> 
> That cannot be quite right, can it? System F has more powerful
> polymorphism than HM.

As I recall HM is along the edge to \lambda^2.


Haskell 98 is typically described in terms of System F_\omega. I've 
heard people say GADTs are a subset of (or equivalent to?) System F_c, 
though I'm not familiar with F_c myself. Like most interesting 
languages, Haskell as a whole doesn't belong on one of the vertices. 
I've yet to see a nice diagram that presents non-vertex languages like 
HM and what arcs can cause one to (often unexpectedly) degenerate into 
another.

-- 
Live well,
~wren