design

::object (struct or class)
	hold shape, size, material properties, and transforms
	might hold some bounding box information or something
	if I want to be able to call some functions on these (like intersect) then I 		should use a class
	maybe merge with ::primitive, might not. They work closely together

::light (probably a struct)
	hold position and color

::primitive
	test for intersections with rays
	more specifically: hold the required geometry primitives, and the equations for
	checking for intersections

::readfile
	parse size into ::scene or ::camera or ::sampler or ::image

::scene
	has an array of objects

::main
	read input	

::sampler
	has an image w x h	I guess

::camera
	has a scene
	has a position in a scene
	has modelview transform function
	creates rays (in the Berkeley spec)
		Ray ray = RayThruPixel(cam, i, j)
	

::raytracer
	find the closest object (how?)
	get the color of a ray using the primitive class
	call yourself recursively to get any subsequent reflections

::film
	Manage the image, store its colors

notes: 
unit tests perhaps
NVidia Optix for speed


The big advantage of a class over a struct is the constructor
typedef Ray vec4?
or struct?
or class?
2 types of rays : shadow rays (is there an intersection at all?) and primary rays
				  (where are the intersections?)
				  I guess primary rays are used for reflective surfaces and shadows
				  for