What should I work on next?

Sat, 09/06/2008 - 00:59 — Boris
Buoyancy controller
41% (9 votes)
Weld joint
32% (7 votes)
A game
5% (1 vote)
This site
0% (0 votes)
The wiki
9% (2 votes)
Shape description format
5% (1 vote)
Deferred filtering
9% (2 votes)
Total votes: 22

Details

I still have some free time left this summer, so I'm considering which of my many idea's to put forward, mainly on Box2D. I'll explain the options:

Buoyancy - I've got an idea to implement buoyancy in Box2D using Monte Carlo integration. Pros: Easy, fast (I hope), can do a variety of fluid effects and shapes, volume based instead of mass based. Cons: Less accurate than cutting up shapes precisely, slight latency in behaviour.

Weld joint - This is a classic joint which just fixes two things together, that Box2D is missing for some reason. Should be easy to implement. As a bonus, you will get an XWeld joint, a vital component in creating certain effects.

A game - I got into programming when young to make games, even if I never actually got on to the game part. But it's never too late. I'd likely start small with some flash game.

The wiki - Box2D's documentation is woeful, aside from the original manual. The wiki is deplorable (I should know, being responsible for it). But it's still the best avenue for improvement, and I still think sufficient work on it would greatly improve takeup rates.

SDF - A simple interchange format for describing 2d shapes. Could be useful. Previous attempts have failed as they try to encapsulate the physics too much, which varies too much in each implementation. I'd draw the line at material properties, and produce a parser for Flash.

Deferred filtering - I have a hunch, that oft-requested, oft-denied feature of being able to filter collisions on the Add event when they occur, is possible to implement despite structural arguments against it. No guarantees tho.