Newton 2.0x Archemedia Open Beta

[Julio Jerez]

you are on linux right?

No, I am mainly on Windows. But I develop also under Linux and MacOSX
Julio, you could add "int shapeID" argument also for HullModifier and CompoundCollision Create functions. Then there will be no need for SetUserID function.

I like that better, it make moe consistent, and also SetUserID can mess up the cache since the id is part of the shape signature now.
I will do that instead and post the fix tonight.

[Julio Jerez]

Ok I added beta 2.06 with the change.

[martinsm]

Great. Thanks for fast fix.
Now 2.06 works as I expected in this aspect (same as 1.53 version). I can now continue upgrading my old project to Newton2.

[Julio Jerez]

Beta 2.06 is ready for all platforms.

[Svenstaro]

The link for http://www.newtondynamics.com/downloads ... .06.tar.gz seems invalid. The correct URL is http://www.newtondynamics.com/downloads ... .06.tar.gz. Please capitalize the name of the file on the server.

[Julio Jerez]

Fixed, Thanks

[bmwop]

Hi,
i have a little problem on Linux with the 2.06 build 32 bit:
When a body leanes the world i use the NewtonBodySetMatrix() to set it into a certain point of the world. But it doesn't work.

I worked in 2.04. But not now.
Did i miss something?

/bmwop

[Julio Jerez]

no you did not.
It is my falt I think I added that bug when fixing the last bug, I commented out that part of the code and forgot to uncomment it and debug it.
I will fix it tonight for 2.07

[Julio Jerez]

Hi,
i have a little problem on Linux with the 2.06 build 32 bit:
When a body leanes the world i use the NewtonBodySetMatrix() to set it into a certain point of the world. But it doesn't work.

Ok I fixed the bug, get beta 2.07

[bmwop]

Ok I fixed the bug, get beta 2.07

Great! Now, it's okay again. Thank you!

[martinsm]

hi, Julio.
I remember asking you one feature request for new Newton version, but that topic/posts have been deleted. You said yes, but it's not in 2.07 version.

Would it be possible to add collision deletion callback? It should be similar as callbacks for world/body/joint deletion. It would help a lot for managed world wrappers (.NET for example).
API could look something like this:

Code: Select all

typedef void (*NewtonCollisionDestructor)(const NewtonCollision* collision);

NEWTON_API void NewtonCollisionSetDestructorCallback(const NewtonCollision* collision, NewtonCollisionDestructor callback);

[Julio Jerez]

Ok I will add it as a world function

NEWTON_API void NewtonWorkdSetCollisionDestructorCallback(const NewtonWorld* world, NewtonCollisionDestructor callback);

[martinsm]

Great, thanks!

[Julio Jerez]

Ok I added the function, you get it withnext update.
I will see if I can find the slowdowng bug before a release.

[rvangaal]

(using 2.07) I had problems before with convex hulls hitting single-sided tree collisions (2.02 or 2.03). Seems better with 2.07 but only if I don't optimized the tree collision.
So if I use NewtonTreeCollisionEndBuild(coll,1); instead of NewtonTreeCollisionEndBuild(coll,0); the problem re-appears (using a car simulation). Once the convex hull of the car hits a single-sided polygon (like a guardrail) it explodes (NaNs). Without the optimize, it also gets a big jerk but stays out of NaN.

Are single-sided triangles forbidden or is it planned to support them in v2.x? (I also have the problem with fences a lot - those are often drawn using 1 polygon, without culling so both sides become visible; from one side you can drive through them).