[PRL] first-class polymorphic values

[Paul A. Steckler]

> Sam answered my question pretty well in this thread. I'm 
> curious about 
> whether there are opportunities for unintended type variable capture 
> when you have a sufficiently sophisticated type language, in 
> particular 
> with polymorphic type application and free type variables.

I'd look at the Weirich boxy types paper, then.

I'm sure you're aware of the issues in Hindley-Milner type 
systems, where you have to be careful about what type 
variables you generalize over.  There may be additional 
subtleties with type variables in the presence of 
first-class polymorphism, but that's not something 
I've studied.

-- Paul