On Tue, Aug 24, 2010 at 11:09:45AM +0200, Csaba Hruska wrote:

I'm currently working on an improved version of bullet physics engine binding. (the binding code is generated by a python script according c++ api. it uses c2hs too) I'd like to support jhc too. The attached code compiles with ghc (ghc --make testBullet.hs) but fails with jhc 0.7.6-3. Here is the error message:

*Typechecking...* *[1 of 4] BulletTypes (.............................................)* *[2 of 4] C2HS (.............................................)* *[3 of 4] BulletMod (......jhc: user error (* *What: failure* *Why: context reduction, no instance for: Jhc.Float.RealFloat Foreign.C.Types.CFloat* *Where: on line 7114 in ./BulletMod.hs* *in the explicitly typed BulletMod.btConcaveShape_setMargin BulletMod.5833_a1* * BulletMod.5834_a2* * = BulletMod.withBt BulletMod.5833_a1 Jhc.Basics.$* * \ BulletMod.5835_a1' ->* * let BulletMod.5836_a2' = C2HS.cFloatConv BulletMod.5834_a2 in* * BulletMod.btConcaveShape_setMargin'_ BulletMod.5835_a1'* * BulletMod.5836_a2'* * Jhc.Monad.>>=* * \ BulletMod.5837_res -> Jhc.Monad.return (Jhc.Basics.())** **{- on line 7114 -}* *Compilation of module: BulletMod)*

Is this my fault or is it a jhc problem?

It's a jhc problem. :) A fairly straightforward one actually, I just never got around to filling out all the instances for the Foreign.C.Types types so it is just a library issue. Thanks for reporting it! John -- John Meacham - ⑆repetae.net⑆john⑈ - http://notanumber.net/