Malcolm Wallace wrote:
(2) I'm hoping that a little extra visibility for the (largely working) current status of nhc98 might encourage some more developers to take nhc98 more seriously, and maybe even start contributing to it. It is good to have more than a single implementation of the Haskell language.
I'll take nhc98 seriously when it can compile GHC, which IIRC is just Rank2Types away (plus writing some nhc98-style .hi files resembling the .hs-boots; the new ghc build-system, being cabal, is a bit friendlier now) I really don't want to try emulating Rank2Types with unsafeCoerce. And it's useful enough that I don't want to deny it to GHC (it had a few uses, last time I looked, that were not obviously removable). But I'm very likely to try, as soon as it becomes possible. Heck, even implementing Rank2Types in a type-unsafe way (implicitly inserting those unsafeCoerces) would be fine with me, if it worked :-) What happened with that alternate type-checker was being developed to possibly replace the tangled nyhc one anyway? nothing? :-) -Isaac