Glynn Clements wrote:
[I'd be surprised if no-one else has encountered this yet, ...]
Well, almost...: :-) http://cvs.haskell.org/cgi-bin/cvsweb.cgi/fptools/libraries/GLUT/Graphics/UI/GLUT/Callbacks/Window.hs.diff?r1=1.5&r2=1.6 The new GLUT binding (including extensive documentation) will be included in the next HOpenGL release, which I *really* hope to get out of the door soon. Apart from that, the GLU API will be much nicer, e.g. tessellation basically looks like "polygon in, triangles/contours/primitives out", without any need for silly callbacks. While I'm at it: I really like to hear opinions and improvements of the API, e.g. the vertex specification stuff like class Vertex a where vertex :: a -> IO () data Vertex2 a = Vertex2 a a instance Vertex (Vertex2 GLshort) where -- complex instance head!!! vertex (Vertex2 x y) = vertex2s x y %fun glVertex2s :: GLshort -> GLshort -> IO () is not Haskell98. How can we do better, i.e. unify 1D/2D/3D/4D vertices with differing component types in pure Haskell98? Cheers, S.