Newton 2.24 Problem with AutoSleep

[mak]

Hi there,

first - my apologies if this is answered elsewhere, I spent a while searching the forums but to no avail.

I've just leapt from 1.53 to 2.24 in order to get the iPhone libraries, but I'm having problems with 2.24 on Windows at the moment. I've taken a project which works just fine under 1.53, and updated the API usage to conform with 2.24. Largely this involved the additional arg passing of -1 for shapeId, and 0 for threadIndex (it's single threaded) . No logic has changed on my side of the fence, and the update was quite smooth.

But, when running my project - everything falls through the "floor" of the room mesh which acts as the container for the scene. The room is constructed from a model by adding each tri to a CollisionTree, and then finally passing that to the NewtonBody to construct it from the NewtonCollision object.

Basically, it seems the room is always asleep/in equilibrium - even when I immediately call NewtonBodySetAutoSleep(body, 0) as soon as I construct it - somehow it gets turned back on (no code of mine does this, I've ripped out all possible calls). I've got debug rendering which colours the mesh based on it's sleep/frozen/autosleep settings, so it's easy to spot what's happening.

My other objects behave just fine against each other - even when allowing AutoSleep - the only difference is that they are created using NewtonCreateConvexHull(...)

Even if I fire a working object at the faces of the room, the room doesn't wake up on contact and the object goes sailing right through it.

Any ideas? I've missed something in the API update? Some new logic path/setting which is needed?

Thanks in advance!

[mak]

Solved myself - for some reason prior to 2.24 adding my custom mesh verts as A, B,C per tri worked - but now with 2.24, I have to reverse it and add C, B, A - and voila, it works.

Phew!

[Julio Jerez]

Oh the faces were in different winding order? that could be, there are many architectural changes between 1.0 and 2.0 change.
but if tha was teh case then the collision shoucl have bit invisible to incomming faces.

I had then is visual studio source control, but aftet I swith to SVN, i chneg compurer few times and I lost those archives.
I has some very older 1.00 version in cd, but not worth to look at because they data from when it only handled Boxes, spheres and collision trees, and has with the bug of the time.

I am glad I have the code in google now, because it much safe and anyone go back to any version, unfortunally 1.53 is history, and even some elray version on 2.00 as well.
no one use newton should be in 1.53