Error in btAxisSweep3Internal, NULL proxy

cyrsep
Posts: 3
Joined: Wed Jun 29, 2011 2:29 pm
Location: France , Paris

Error in btAxisSweep3Internal, NULL proxy

Post by cyrsep »

Hi there !

I' m having errors in btAxisSweep3Internal thing.
The error changes from time to time but always occurs in this part of the code.
it's mainly because of the proxy pointer set to NULL in the setAabb() function but i cant figure out why.

I've tried to increase the number of max handles but all i got is a full ram and the error still occurs

Any ideas ?

USING : Bullet 2.75
User avatar
Erwin Coumans
Site Admin
Posts: 4221
Joined: Sun Jun 26, 2005 6:43 pm
Location: California, USA

Re: Error in btAxisSweep3Internal, NULL proxy

Post by Erwin Coumans »

I haven't seen such problem before. Could it be that some data is not properly initialized.
A common mistake is to not initialize the btTransform properly (use transform.setIdentity() etc). Uninitialized collision shapes could be another issue.

You might want to try updating to Bullet 2.78 and see if it is fixed.
cyrsep
Posts: 3
Joined: Wed Jun 29, 2011 2:29 pm
Location: France , Paris

Re: Error in btAxisSweep3Internal, NULL proxy

Post by cyrsep »

Thanks for the quick reply.

I've checked all the code. The motion state and the collision shape are both initialised.

I've also noticed that it happens mainly (as it is a random error) when I set the Collision Flag to 4 (CF_NO_CONTACT_RESPONSE)
When I remove this option, the error seem to disappear. The thing is that I need this flag so the rigid Bodies I create don't interact with each other.
Still, I need to use the rigid body as my object has a mass and receive a "quick" at the beginning of the animation.

I'm not sure I can switch to bullet 2.78 as the whole project began with 2.75. I don't want to change all the work done in the past by my colleagues ( as long as I can ).
cyrsep
Posts: 3
Joined: Wed Jun 29, 2011 2:29 pm
Location: France , Paris

Re: Error in btAxisSweep3Internal, NULL proxy

Post by cyrsep »

Just to say that the error came from a bad thread control, which causes delete on rigidbodies while the world is calculating a step.

Still, I do have some errors when I have multiple objects (in my case 1000). Do you have any idea to make bullet work faster and without errors or should i
tell my boss we've reached the limits ? :?