4 Feb
2006
4 Feb
'06
8:53 p.m.
On Thu, Feb 02, 2006 at 03:53:11PM +0000, Henrik Nilsson wrote:
What about the "FD Conditions" as defined in the paper by Sulzman, Duck, Peyton-Jones and Stuckey you referred to? (Mark Jones original conditions + a so called "bound variable condition"). If those conditions are imposed, one gets sound, complete, and decidable type inference according to the said paper.
I had a quick look at the monadic libraries and a vector space class in Yampa, and my impression is that the FDs used there would be compatible with the FD conditions. But I emphasize that's just an impression.
Sadly not for monad transformers -- see under Details on the FunctionalDependencies wiki page.