freeze bug at NewtonCollisionUpdate() [bugfixed]

Hi,

I found a bug in a new function of Newton 2.
In special situations there is a freeze (deadlog) by calling NewtonCollisionUpdate().
(In Mogrenewt: World.CollisionUpdate())

Now I wrote a test application which reproduces the problem.
The developer of the Newton physics library wants fix the problem,
Now my test application runs on his computer,
BUT he can't debug the linked Newton library, because the test application is C#.

He has no C# experience.
And I have less knowledge about C++.

We try to find a way how to debug.
The alternative would be to port the source code of my test application.
Maybe somebody of you would like to do?

Look here for more information:
viewtopic.php?f=8&t=10113&start=0

(yes, it's a cross post, but which normal user looks to the Mogre section )

Ok, now the author of Newton found a way how to debug his linked library.
So DON'T PORT the application. There is not anymore a need.

Now the bug is fixed
I made several tests and everything is fine.
And for small vs. huge objects a contact still will be detected.

Thanks to the author of Newton - who spend about one full day to catch the problem.
(Most of the time was to get run the test application on his computer and find out how to debug the linked library.)

I I found it, I had to go to a serius of hooks setting security and debug options.
I do not know what flag did it by now I can debug.
I still can not step for the Newton SDK but it is OK, i can debug like this.

any way I found the bug.

you have very huge bodies that are colliding with small bodies,
the bug happens when build the Mincosky hull for such pair.
the resulting hull is soo flat that 32 bit floats lose presistion.
I have a fallback precedure but even the fallback fail.

I am fixing it now. stand by