Name:
	Ben Lynn
(captain/team name)

Email:
	ben@staff.cs.usyd.edu.au

Address:
	40 Second Ave.
	Lane Cove NSW 2066
	Australia

Language:
	Eiffel (SmallEiffel -0.76beta#4)

Comments:
	Tier 3

	I'm not very good at functional languages (I could really use a
	book on the subject :) ), which is why I chose to use Eiffel - a
	clean object-oriented language that can interface to C fairly
	easily if the need arises.

History:
	Very minor change: now only allows characters 32 to 126 in strings
	(except for " of course).


	Minor optimizations, bugfixes (mostly special cases I left out
	before). I dare not attempt any major optimizations with the deadline
	so near.


	Made several optimizations. Discovered yet another major bug after
	reading about the dice.ppm update: my shadow rays weren't taking
	into account the distance to the light source (they assumed all
	lights were at infinity). Plan to optimize further.


	Added bounding boxes, which speeded things up a little. I had the
	same bug as the sample implementation on reflect.gml (I was
	returning empty lists, rather than a span starting from 0 and ending
	at the point of intersection).
	A small numerical problem involving the CSG difference operator
	was fixed.
	Fixed cones and cylinders, the code left out a few cases.
	More optimizations to go, I'll submit first just in case.


	Cone shadow no longer has acne. Slightly optimized CSG operations.


	Found out about Task misprint. Changed lighting model (d cubed to
	d squared).


	Tier 3 complete. Renders all sample files, though my version of the
	dice look a bit different.  A few more bug fixes - I was previously
	recording distances (from the viewer) incorrectly! Hopefully not my
	last submission. I'm going to try to fix a few minor problems and
	optimize.


	More bug fixes; mostly texture mapping problems.

	I've now done Tier 2, though cones can have bad shadows
	due to numerical inaccuracies - I'll try to fix it.
	Fixed some bugs, and the code is cleaner.


	Not intended to be final submission. (I was glad to get Tier 1 over
	and done with, I submitted mainly to feel good.) Hasn't been
	thoroughly tested yet, though it agrees with the samples.

	Code is messy. Started off relatively neat, and some bits I think
	are well-designed, but unfortunately I haven't had much experience
	with raytracers and the code quality deteriorated.

	I didn't start the contest on time due to the time difference. :(
