Static mesh bug in build 2565

[Bird]

Hi Julio,

I'm seeing strange behavior in today's build 2565 with a simple static collision tree mesh(the ground mesh). Here's the serialize scene file.
http://www.hurleyworks.com/downloads/BuggyStaticMeshv2565.zip

If I change the ground mesh to a box collision shape instead of a collision tree then the simulation works as expected ... the cube slides smoothly down the ground.

-Bird

[Sweenie]

The box seems to collide with the triangle edge.

[Julio Jerez]

NP that is not normal, that's a huge bug. I know exactle what it is, and I thought I had the fixed.
it has to do with the welded edge, there most be a bug wit the normal rotation to the voroni region of the collidng edge face when it is hit the first time.

The problem is that as the box slide down and it his the edge the first time, the minimal geometrical gistance that resolve is selcetec as the direction if the contact.
in that case it is a contact that is almost tanget to the face, instead fo being normal face.
In core 200 these contact direction were part of the shape face calculations, but in core 300 is a posprosess after all contact are calculated.
there most be a bad error there, must be a bad error there.

checking it now.

[Julio Jerez]

Al right that was a stupic copy and past mistake.
I check in the fix. However thsi si no teh definite solution for polgion edeg contact.

I am adding that modification now, and it will be check in maybe tomorrow.
This is importnat and this bug reminded me of that part that I posponed but that nwo is tiem to complete.

if you sync it shoul work now. do no worrie if it falke in soem oteh place I am fixing the part now.
please Keep those bugs comming, They can only make the code stronger