NewtonCollisionDestructor() issue

[thedmd]

Hello Julio,

I found NewtonCollisionDestructor() a little problematic. I'm working on managed version of Newton, but there are some problems with managing object destruction time. In this case NewtonCollision objects exacly. Callback NewtonCollisionDestructor() is set in NewtonWorld and it is called just with one argument, pointer to NewtonCollision. Becouse of that I cannot invoke proper managed callback.
In practise I can call callback for NewtonWorld which created NewtonCollision which is safe until NewtonCollisionMakeUnique() is called. After that I need to track NewtonWorld object.

Instead of making a fancy workarounds and 'current' world tracking I wish to ask you about doing following modification in API:

Code: Select all

typedef void (*NewtonCollisionDestructor) (const NewtonWorld* world, const NewtonCollision* collision);

[Julio Jerez]

It is not a problem but you most call
void NewtonReleaseCollision (const NewtonWorld* newtonWorld, const NewtonCollision* collision)

from a world, isn't that sufficient?

[JernejL]

The problem is, he uses callback to handle actual validation of destruction and there he only knows collision handle and not newton world, adding a newtoncollisiongetworld should solve the problem.

[Julio Jerez]

I see,
Ok I add the extra argument

[thedmd]

NewtonCollisionGetWorld() in fact solves the problem, but there is not actual world bounded to NewtonCollision itself. Referenced World is only known while calling NewtonReleaseCollision(). Becouse of that I think additional parameter in callback is better solution.

Thank's for (quick) support.