Re: [Supertuxkart-devel] Physics improvements/suggestions

[David <inf...@gm...>]

I actually have some code this time :D -- although it is still more of a 
test than an actual improvement at the moment. I thought I'd post it so 
that you know I'm still slowly working on this and to maybe get some 
more advice from hiker.

> I just realised something:
>> Currently the src/physics/btkart.cpp code has 2 hacks that are purposely 
>> put in there, but lead to unwanted side effects.
>>     
> There is actually a third important change included: the bullet skidding 
> code is removed. So as recommended before: start with a clean raycast 
> vehicle from bullet, and don't base anything on btKart.
>   
Okay. I started with bullets btRaycastVehicle code and started patching 
it to look like the current btKart code. There was actually quite a few 
small patches to add and they all actually make a lot of sense (albeit 
with some side extra effects for one in particular I mentioned before). 
If by 'skidding' you mean 'sliding,' I set a boolean to switch this 
on/off (and found it better off :P).

I then added code to check for changes in slope between frames so that 
the kart doesn't suddenly find itself on a different slope. The patched 
code (although a little messy) should do this reasonably well and give 
clear output for when this happens.

The only problem is that this doesn't appear to be as bad a problem as I 
thought it would be :S

I want the karts to be able to go up smooth, curved ramps at high 
velocity (this would have other benefits such as making it easier to 
travel on upwards gradients, or making sure the kart does not pass 
through certain sloped surfaces and get rescued at high speeds or being 
pushed by an explosion).

Could it be the wheels go up the ramps fine, but the frame of the kart, 
which is larger, collides with the angled surface ahead? How do I switch 
on bullet debugging to see the actual shapes?