virtual void addContactPoint(const btVector3& normalOnBInWorld, const btVector3& pointInWorldOrg, btScalar depthOrg) 
			{
				m_reportedDistance = depthOrg;
				m_reportedNormalOnWorld = normalOnBInWorld;
				
				btVector3 adjustedPointB = pointInWorldOrg - normalOnBInWorld*m_marginOnB;
				m_reportedDistance = depthOrg+(m_marginOnA+m_marginOnB);
				if (m_reportedDistance<0.f)
				{
					m_foundResult = true;					
				}
				m_originalResult->addContactPoint(normalOnBInWorld, adjustedPointB, m_reportedDistance);
			}
bool btPolyhedralContactClipping::findSeparatingAxis(	const btConvexPolyhedron& hullA, const btConvexPolyhedron& hullB, const btTransform& transA, const btTransform& transB, btVector3& sep, btDiscreteCollisionDetectorInterface::Result& resultOut)
{
	gActualSATPairTests++;

//#ifdef TEST_INTERNAL_OBJECTS
	const btVector3 c0 = transA * hullA.m_localCenter;
	const btVector3 c1 = transB * hullB.m_localCenter;
	const btVector3 DeltaC2 = c0 - c1;
//#endif

	btScalar dmin = FLT_MAX;
	int curPlaneTests=0;

	int numFacesA = hullA.m_faces.size();
	// Test normals from hullA
	for(int i=0;i<numFacesA;i++)
	{
		const btVector3 Normal(hullA.m_faces[i].m_plane[0], hullA.m_faces[i].m_plane[1], hullA.m_faces[i].m_plane[2]);
		btVector3 faceANormalWS = transA.getBasis() * Normal;
		if (DeltaC2.dot(faceANormalWS)<0)
			faceANormalWS*=-1.f;

		curPlaneTests++;
#ifdef TEST_INTERNAL_OBJECTS
		gExpectedNbTests++;
		if(gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, faceANormalWS, hullA, hullB, dmin))
			continue;
		gActualNbTests++;
#endif

		btScalar d;
		btVector3 wA, wB;
		if(!TestSepAxis( hullA, hullB, transA, transB, faceANormalWS, d, wA, wB))
			return false;

		if(d<dmin)
		{
			dmin = d;
			sep = faceANormalWS;
		}
	}

	int numFacesB = hullB.m_faces.size();
	// Test normals from hullB
	for(int i=0;i<numFacesB;i++)
	{
		const btVector3 Normal(hullB.m_faces[i].m_plane[0], hullB.m_faces[i].m_plane[1], hullB.m_faces[i].m_plane[2]);
		btVector3 WorldNormal = transB.getBasis() * Normal;
		if (DeltaC2.dot(WorldNormal)<0)
			WorldNormal *=-1.f;

		curPlaneTests++;
#ifdef TEST_INTERNAL_OBJECTS
		gExpectedNbTests++;
		if(gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, WorldNormal, hullA, hullB, dmin))
			continue;
		gActualNbTests++;
#endif

		btScalar d;
		btVector3 wA, wB;
		if(!TestSepAxis(hullA, hullB, transA, transB, WorldNormal, d,wA, wB))
			return false;

		if(d<dmin)
		{
			dmin = d;
			sep = WorldNormal;
		}
	}

	btVector3 edgeAstart, edgeAend, edgeBstart, edgeBend;
	int edgeA=-1;
	int edgeB=-1;
	btVector3 worldEdgeA;
	btVector3 worldEdgeB;
	btVector3 witnessPointA, witnessPointB;
	

	int curEdgeEdge = 0;
	// Test edges
	for(int e0=0;e0<hullA.m_uniqueEdges.size();e0++)
	{
		const btVector3 edge0 = hullA.m_uniqueEdges[e0];
		const btVector3 WorldEdge0 = transA.getBasis() * edge0;
		for(int e1=0;e1<hullB.m_uniqueEdges.size();e1++)
		{
			const btVector3 edge1 = hullB.m_uniqueEdges[e1];
			const btVector3 WorldEdge1 = transB.getBasis() * edge1;

			btVector3 Cross = WorldEdge0.cross(WorldEdge1);
			curEdgeEdge++;
			if(!IsAlmostZero(Cross))
			{
				Cross = Cross.normalize();
				if (DeltaC2.dot(Cross)<0)
					Cross *= -1.f;


#ifdef TEST_INTERNAL_OBJECTS
				gExpectedNbTests++;
				if(gUseInternalObject && !TestInternalObjects(transA, transB, DeltaC2, Cross, hullA, hullB, dmin))
					continue;
				gActualNbTests++;
#endif

				btScalar dist;
				btVector3 wA, wB;
				if(!TestSepAxis( hullA, hullB, transA, transB, Cross, dist, wA, wB))
					return false;

				if(dist<dmin)
				{
					dmin = dist;
					sep = Cross;
					edgeA=e0;
					edgeB=e1;
					worldEdgeA = WorldEdge0;
					worldEdgeB = WorldEdge1;
					witnessPointA=wA;
					witnessPointB=wB;
				}
			}
		}

	}

	if (edgeA>=0&&edgeB>=0)
	{
//		printf("edge-edge\n");
		//add an edge-edge contact

		btVector3 ptsVector;
		btVector3 offsetA;
		btVector3 offsetB;
		btScalar tA;
		btScalar tB;

		btVector3 translation = witnessPointB-witnessPointA;

		btVector3 dirA = worldEdgeA;
		btVector3 dirB = worldEdgeB;
		
		btScalar hlenB = 1e30f;
		btScalar hlenA = 1e30f;

		btSegmentsClosestPoints(ptsVector, offsetA, offsetB, tA, tB,
			translation,
			dirA, hlenA,
			dirB, hlenB);

		btScalar nlSqrt = ptsVector.length2();
		if (nlSqrt>SIMD_EPSILON)
		{
			btScalar nl = btSqrt(nlSqrt);
			ptsVector *= 1.f/nl;
			if (ptsVector.dot(DeltaC2)<0.f)
			{
				ptsVector*=-1.f;
			}
			btVector3 ptOnB = witnessPointB + offsetB;
			btScalar distance = nl;
			resultOut.addContactPoint(ptsVector, ptOnB,-distance);
		}

	}


	if((DeltaC2.dot(sep))<0.0f)
		sep = -sep;

	return true;
}
void	btPolyhedralContactClipping::clipFaceAgainstHull(const btVector3& separatingNormal, const btConvexPolyhedron& hullA,  const btTransform& transA, btVertexArray& worldVertsB1, const btScalar minDist, btScalar maxDist,btDiscreteCollisionDetectorInterface::Result& resultOut)
{
	btVertexArray worldVertsB2;
	btVertexArray* pVtxIn = &worldVertsB1;
	btVertexArray* pVtxOut = &worldVertsB2;
	pVtxOut->reserve(pVtxIn->size());

	int closestFaceA=-1;
	{
		btScalar dmin = FLT_MAX;
		for(int face=0;face<hullA.m_faces.size();face++)
		{
			const btVector3 Normal(hullA.m_faces[face].m_plane[0], hullA.m_faces[face].m_plane[1], hullA.m_faces[face].m_plane[2]);
			const btVector3 faceANormalWS = transA.getBasis() * Normal;
		
			btScalar d = faceANormalWS.dot(separatingNormal);
			if (d < dmin)
			{
				dmin = d;
				closestFaceA = face;
			}
		}
	}
	if (closestFaceA<0)
		return;

	const btFace& polyA = hullA.m_faces[closestFaceA];

		// clip polygon to back of planes of all faces of hull A that are adjacent to witness face
	int numContacts = pVtxIn->size();
	int numVerticesA = polyA.m_indices.size();
	for(int e0=0;e0<numVerticesA;e0++)
	{
		const btVector3& a = hullA.m_vertices[polyA.m_indices[e0]];
		const btVector3& b = hullA.m_vertices[polyA.m_indices[(e0+1)%numVerticesA]];
		const btVector3 edge0 = a - b;
		const btVector3 WorldEdge0 = transA.getBasis() * edge0;
		btVector3 worldPlaneAnormal1 = transA.getBasis()* btVector3(polyA.m_plane[0],polyA.m_plane[1],polyA.m_plane[2]);

		btVector3 planeNormalWS1 = -WorldEdge0.cross(worldPlaneAnormal1);//.cross(WorldEdge0);
		btVector3 worldA1 = transA*a;
		btScalar planeEqWS1 = -worldA1.dot(planeNormalWS1);
		
//int otherFace=0;
#ifdef BLA1
		int otherFace = polyA.m_connectedFaces[e0];
		btVector3 localPlaneNormal (hullA.m_faces[otherFace].m_plane[0],hullA.m_faces[otherFace].m_plane[1],hullA.m_faces[otherFace].m_plane[2]);
		btScalar localPlaneEq = hullA.m_faces[otherFace].m_plane[3];

		btVector3 planeNormalWS = transA.getBasis()*localPlaneNormal;
		btScalar planeEqWS=localPlaneEq-planeNormalWS.dot(transA.getOrigin());
#else 
		btVector3 planeNormalWS = planeNormalWS1;
		btScalar planeEqWS=planeEqWS1;
		
#endif
		//clip face

		clipFace(*pVtxIn, *pVtxOut,planeNormalWS,planeEqWS);
		btSwap(pVtxIn,pVtxOut);
		pVtxOut->resize(0);
	}



//#define ONLY_REPORT_DEEPEST_POINT

	btVector3 point;
	

	// only keep points that are behind the witness face
	{
		btVector3 localPlaneNormal (polyA.m_plane[0],polyA.m_plane[1],polyA.m_plane[2]);
		btScalar localPlaneEq = polyA.m_plane[3];
		btVector3 planeNormalWS = transA.getBasis()*localPlaneNormal;
		btScalar planeEqWS=localPlaneEq-planeNormalWS.dot(transA.getOrigin());
		for (int i=0;i<pVtxIn->size();i++)
		{
			
			btScalar depth = planeNormalWS.dot(pVtxIn->at(i))+planeEqWS;
			if (depth <=minDist)
			{
//				printf("clamped: depth=%f to minDist=%f\n",depth,minDist);
				depth = minDist;
			}

			if (depth <=maxDist)
			{
				btVector3 point = pVtxIn->at(i);
#ifdef ONLY_REPORT_DEEPEST_POINT
				curMaxDist = depth;
#else
#if 0
				if (depth<-3)
				{
					printf("error in btPolyhedralContactClipping depth = %f\n", depth);
					printf("likely wrong separatingNormal passed in\n");
				} 
#endif				
				resultOut.addContactPoint(separatingNormal,point,depth);
#endif
			}
		}
	}
#ifdef ONLY_REPORT_DEEPEST_POINT
	if (curMaxDist<maxDist)
	{
		resultOut.addContactPoint(separatingNormal,point,curMaxDist);
	}
#endif //ONLY_REPORT_DEEPEST_POINT

}
int dBoxBox2 (const btVector3& p1, const dMatrix3 R1,
	     const btVector3& side1, const btVector3& p2,
	     const dMatrix3 R2, const btVector3& side2,
	     btVector3& normal, btScalar *depth, int *return_code,
		 int maxc, dContactGeom * /*contact*/, int /*skip*/,btDiscreteCollisionDetectorInterface::Result& output)
{
  const btScalar fudge_factor = btScalar(1.05);
  btVector3 p,pp,normalC(0.f,0.f,0.f);
  const btScalar *normalR = 0;
  btScalar A[3],B[3],R11,R12,R13,R21,R22,R23,R31,R32,R33,
    Q11,Q12,Q13,Q21,Q22,Q23,Q31,Q32,Q33,s,s2,l;
  int i,j,invert_normal,code;

  // get vector from centers of box 1 to box 2, relative to box 1
  p = p2 - p1;
  dMULTIPLY1_331 (pp,R1,p);		// get pp = p relative to body 1

  // get side lengths / 2
  A[0] = side1[0]*btScalar(0.5);
  A[1] = side1[1]*btScalar(0.5);
  A[2] = side1[2]*btScalar(0.5);
  B[0] = side2[0]*btScalar(0.5);
  B[1] = side2[1]*btScalar(0.5);
  B[2] = side2[2]*btScalar(0.5);

  // Rij is R1'*R2, i.e. the relative rotation between R1 and R2
  R11 = dDOT44(R1+0,R2+0); R12 = dDOT44(R1+0,R2+1); R13 = dDOT44(R1+0,R2+2);
  R21 = dDOT44(R1+1,R2+0); R22 = dDOT44(R1+1,R2+1); R23 = dDOT44(R1+1,R2+2);
  R31 = dDOT44(R1+2,R2+0); R32 = dDOT44(R1+2,R2+1); R33 = dDOT44(R1+2,R2+2);

  Q11 = btFabs(R11); Q12 = btFabs(R12); Q13 = btFabs(R13);
  Q21 = btFabs(R21); Q22 = btFabs(R22); Q23 = btFabs(R23);
  Q31 = btFabs(R31); Q32 = btFabs(R32); Q33 = btFabs(R33);

  // for all 15 possible separating axes:
  //   * see if the axis separates the boxes. if so, return 0.
  //   * find the depth of the penetration along the separating axis (s2)
  //   * if this is the largest depth so far, record it.
  // the normal vector will be set to the separating axis with the smallest
  // depth. note: normalR is set to point to a column of R1 or R2 if that is
  // the smallest depth normal so far. otherwise normalR is 0 and normalC is
  // set to a vector relative to body 1. invert_normal is 1 if the sign of
  // the normal should be flipped.

#define TST(expr1,expr2,norm,cc) \
  s2 = btFabs(expr1) - (expr2); \
  if (s2 > 0) return 0; \
  if (s2 > s) { \
    s = s2; \
    normalR = norm; \
    invert_normal = ((expr1) < 0); \
    code = (cc); \
  }

  s = -dInfinity;
  invert_normal = 0;
  code = 0;

  // separating axis = u1,u2,u3
  TST (pp[0],(A[0] + B[0]*Q11 + B[1]*Q12 + B[2]*Q13),R1+0,1);
  TST (pp[1],(A[1] + B[0]*Q21 + B[1]*Q22 + B[2]*Q23),R1+1,2);
  TST (pp[2],(A[2] + B[0]*Q31 + B[1]*Q32 + B[2]*Q33),R1+2,3);

  // separating axis = v1,v2,v3
  TST (dDOT41(R2+0,p),(A[0]*Q11 + A[1]*Q21 + A[2]*Q31 + B[0]),R2+0,4);
  TST (dDOT41(R2+1,p),(A[0]*Q12 + A[1]*Q22 + A[2]*Q32 + B[1]),R2+1,5);
  TST (dDOT41(R2+2,p),(A[0]*Q13 + A[1]*Q23 + A[2]*Q33 + B[2]),R2+2,6);

  // note: cross product axes need to be scaled when s is computed.
  // normal (n1,n2,n3) is relative to box 1.
#undef TST
#define TST(expr1,expr2,n1,n2,n3,cc) \
  s2 = btFabs(expr1) - (expr2); \
  if (s2 > SIMD_EPSILON) return 0; \
  l = btSqrt((n1)*(n1) + (n2)*(n2) + (n3)*(n3)); \
  if (l > SIMD_EPSILON) { \
    s2 /= l; \
    if (s2*fudge_factor > s) { \
      s = s2; \
      normalR = 0; \
      normalC[0] = (n1)/l; normalC[1] = (n2)/l; normalC[2] = (n3)/l; \
      invert_normal = ((expr1) < 0); \
      code = (cc); \
    } \
  }

  btScalar fudge2 (1.0e-5f);

  Q11 += fudge2;
  Q12 += fudge2;
  Q13 += fudge2;

  Q21 += fudge2;
  Q22 += fudge2;
  Q23 += fudge2;

  Q31 += fudge2;
  Q32 += fudge2;
  Q33 += fudge2;

  // separating axis = u1 x (v1,v2,v3)
  TST(pp[2]*R21-pp[1]*R31,(A[1]*Q31+A[2]*Q21+B[1]*Q13+B[2]*Q12),0,-R31,R21,7);
  TST(pp[2]*R22-pp[1]*R32,(A[1]*Q32+A[2]*Q22+B[0]*Q13+B[2]*Q11),0,-R32,R22,8);
  TST(pp[2]*R23-pp[1]*R33,(A[1]*Q33+A[2]*Q23+B[0]*Q12+B[1]*Q11),0,-R33,R23,9);

  // separating axis = u2 x (v1,v2,v3)
  TST(pp[0]*R31-pp[2]*R11,(A[0]*Q31+A[2]*Q11+B[1]*Q23+B[2]*Q22),R31,0,-R11,10);
  TST(pp[0]*R32-pp[2]*R12,(A[0]*Q32+A[2]*Q12+B[0]*Q23+B[2]*Q21),R32,0,-R12,11);
  TST(pp[0]*R33-pp[2]*R13,(A[0]*Q33+A[2]*Q13+B[0]*Q22+B[1]*Q21),R33,0,-R13,12);

  // separating axis = u3 x (v1,v2,v3)
  TST(pp[1]*R11-pp[0]*R21,(A[0]*Q21+A[1]*Q11+B[1]*Q33+B[2]*Q32),-R21,R11,0,13);
  TST(pp[1]*R12-pp[0]*R22,(A[0]*Q22+A[1]*Q12+B[0]*Q33+B[2]*Q31),-R22,R12,0,14);
  TST(pp[1]*R13-pp[0]*R23,(A[0]*Q23+A[1]*Q13+B[0]*Q32+B[1]*Q31),-R23,R13,0,15);

#undef TST

  if (!code) return 0;

  // if we get to this point, the boxes interpenetrate. compute the normal
  // in global coordinates.
  if (normalR) {
    normal[0] = normalR[0];
    normal[1] = normalR[4];
    normal[2] = normalR[8];
  }
  else {
    dMULTIPLY0_331 (normal,R1,normalC);
  }
  if (invert_normal) {
    normal[0] = -normal[0];
    normal[1] = -normal[1];
    normal[2] = -normal[2];
  }
  *depth = -s;

  // compute contact point(s)

  if (code > 6) {
    // an edge from box 1 touches an edge from box 2.
    // find a point pa on the intersecting edge of box 1
    btVector3 pa;
    btScalar sign;
    for (i=0; i<3; i++) pa[i] = p1[i];
    for (j=0; j<3; j++) {
      sign = (dDOT14(normal,R1+j) > 0) ? btScalar(1.0) : btScalar(-1.0);
      for (i=0; i<3; i++) pa[i] += sign * A[j] * R1[i*4+j];
    }

    // find a point pb on the intersecting edge of box 2
    btVector3 pb;
    for (i=0; i<3; i++) pb[i] = p2[i];
    for (j=0; j<3; j++) {
      sign = (dDOT14(normal,R2+j) > 0) ? btScalar(-1.0) : btScalar(1.0);
      for (i=0; i<3; i++) pb[i] += sign * B[j] * R2[i*4+j];
    }

    btScalar alpha,beta;
    btVector3 ua,ub;
    for (i=0; i<3; i++) ua[i] = R1[((code)-7)/3 + i*4];
    for (i=0; i<3; i++) ub[i] = R2[((code)-7)%3 + i*4];

    dLineClosestApproach (pa,ua,pb,ub,&alpha,&beta);
    for (i=0; i<3; i++) pa[i] += ua[i]*alpha;
    for (i=0; i<3; i++) pb[i] += ub[i]*beta;

	{
		
		//contact[0].pos[i] = btScalar(0.5)*(pa[i]+pb[i]);
		//contact[0].depth = *depth;
		btVector3 pointInWorld;

#ifdef USE_CENTER_POINT
	    for (i=0; i<3; i++) 
			pointInWorld[i] = (pa[i]+pb[i])*btScalar(0.5);
		output.addContactPoint(-normal,pointInWorld,-*depth);
#else
		output.addContactPoint(-normal,pb,-*depth);

#endif //
		*return_code = code;
	}
    return 1;
  }

  // okay, we have a face-something intersection (because the separating
  // axis is perpendicular to a face). define face 'a' to be the reference
  // face (i.e. the normal vector is perpendicular to this) and face 'b' to be
  // the incident face (the closest face of the other box).

  const btScalar *Ra,*Rb,*pa,*pb,*Sa,*Sb;
  if (code <= 3) {
    Ra = R1;
    Rb = R2;
    pa = p1;
    pb = p2;
    Sa = A;
    Sb = B;
  }
  else {
    Ra = R2;
    Rb = R1;
    pa = p2;
    pb = p1;
    Sa = B;
    Sb = A;
  }

  // nr = normal vector of reference face dotted with axes of incident box.
  // anr = absolute values of nr.
  btVector3 normal2,nr,anr;
  if (code <= 3) {
    normal2[0] = normal[0];
    normal2[1] = normal[1];
    normal2[2] = normal[2];
  }
  else {
    normal2[0] = -normal[0];
    normal2[1] = -normal[1];
    normal2[2] = -normal[2];
  }
  dMULTIPLY1_331 (nr,Rb,normal2);
  anr[0] = btFabs (nr[0]);
  anr[1] = btFabs (nr[1]);
  anr[2] = btFabs (nr[2]);

  // find the largest compontent of anr: this corresponds to the normal
  // for the indident face. the other axis numbers of the indicent face
  // are stored in a1,a2.
  int lanr,a1,a2;
  if (anr[1] > anr[0]) {
    if (anr[1] > anr[2]) {
      a1 = 0;
      lanr = 1;
      a2 = 2;
    }
    else {
      a1 = 0;
      a2 = 1;
      lanr = 2;
    }
  }
  else {
    if (anr[0] > anr[2]) {
      lanr = 0;
      a1 = 1;
      a2 = 2;
    }
    else {
      a1 = 0;
      a2 = 1;
      lanr = 2;
    }
  }

  // compute center point of incident face, in reference-face coordinates
  btVector3 center;
  if (nr[lanr] < 0) {
    for (i=0; i<3; i++) center[i] = pb[i] - pa[i] + Sb[lanr] * Rb[i*4+lanr];
  }
  else {
    for (i=0; i<3; i++) center[i] = pb[i] - pa[i] - Sb[lanr] * Rb[i*4+lanr];
  }

  // find the normal and non-normal axis numbers of the reference box
  int codeN,code1,code2;
  if (code <= 3) codeN = code-1; else codeN = code-4;
  if (codeN==0) {
    code1 = 1;
    code2 = 2;
  }
  else if (codeN==1) {
    code1 = 0;
    code2 = 2;
  }
  else {
    code1 = 0;
    code2 = 1;
  }

  // find the four corners of the incident face, in reference-face coordinates
  btScalar quad[8];	// 2D coordinate of incident face (x,y pairs)
  btScalar c1,c2,m11,m12,m21,m22;
  c1 = dDOT14 (center,Ra+code1);
  c2 = dDOT14 (center,Ra+code2);
  // optimize this? - we have already computed this data above, but it is not
  // stored in an easy-to-index format. for now it's quicker just to recompute
  // the four dot products.
  m11 = dDOT44 (Ra+code1,Rb+a1);
  m12 = dDOT44 (Ra+code1,Rb+a2);
  m21 = dDOT44 (Ra+code2,Rb+a1);
  m22 = dDOT44 (Ra+code2,Rb+a2);
  {
    btScalar k1 = m11*Sb[a1];
    btScalar k2 = m21*Sb[a1];
    btScalar k3 = m12*Sb[a2];
    btScalar k4 = m22*Sb[a2];
    quad[0] = c1 - k1 - k3;
    quad[1] = c2 - k2 - k4;
    quad[2] = c1 - k1 + k3;
    quad[3] = c2 - k2 + k4;
    quad[4] = c1 + k1 + k3;
    quad[5] = c2 + k2 + k4;
    quad[6] = c1 + k1 - k3;
    quad[7] = c2 + k2 - k4;
  }

  // find the size of the reference face
  btScalar rect[2];
  rect[0] = Sa[code1];
  rect[1] = Sa[code2];

  // intersect the incident and reference faces
  btScalar ret[16];
  int n = intersectRectQuad2 (rect,quad,ret);
  if (n < 1) return 0;		// this should never happen

  // convert the intersection points into reference-face coordinates,
  // and compute the contact position and depth for each point. only keep
  // those points that have a positive (penetrating) depth. delete points in
  // the 'ret' array as necessary so that 'point' and 'ret' correspond.
  btScalar point[3*8];		// penetrating contact points
  btScalar dep[8];			// depths for those points
  btScalar det1 = 1.f/(m11*m22 - m12*m21);
  m11 *= det1;
  m12 *= det1;
  m21 *= det1;
  m22 *= det1;
  int cnum = 0;			// number of penetrating contact points found
  for (j=0; j < n; j++) {
    btScalar k1 =  m22*(ret[j*2]-c1) - m12*(ret[j*2+1]-c2);
    btScalar k2 = -m21*(ret[j*2]-c1) + m11*(ret[j*2+1]-c2);
    for (i=0; i<3; i++) point[cnum*3+i] =
			  center[i] + k1*Rb[i*4+a1] + k2*Rb[i*4+a2];
    dep[cnum] = Sa[codeN] - dDOT(normal2,point+cnum*3);
    if (dep[cnum] >= 0) {
      ret[cnum*2] = ret[j*2];
      ret[cnum*2+1] = ret[j*2+1];
      cnum++;
    }
  }
  if (cnum < 1) return 0;	// this should never happen

  // we can't generate more contacts than we actually have
  if (maxc > cnum) maxc = cnum;
  if (maxc < 1) maxc = 1;

  if (cnum <= maxc) {

	  if (code<4) 
	  {
    // we have less contacts than we need, so we use them all
    for (j=0; j < cnum; j++) 
	{
		btVector3 pointInWorld;
		for (i=0; i<3; i++) 
			pointInWorld[i] = point[j*3+i] + pa[i];
		output.addContactPoint(-normal,pointInWorld,-dep[j]);

    }
	  } else
	  {
		  // we have less contacts than we need, so we use them all
		for (j=0; j < cnum; j++) 
		{
			btVector3 pointInWorld;
			for (i=0; i<3; i++) 
				pointInWorld[i] = point[j*3+i] + pa[i]-normal[i]*dep[j];
				//pointInWorld[i] = point[j*3+i] + pa[i];
			output.addContactPoint(-normal,pointInWorld,-dep[j]);
		}
	  }
  }
  else {
    // we have more contacts than are wanted, some of them must be culled.
    // find the deepest point, it is always the first contact.
    int i1 = 0;
    btScalar maxdepth = dep[0];
    for (i=1; i<cnum; i++) {
      if (dep[i] > maxdepth) {
	maxdepth = dep[i];
	i1 = i;
      }
    }

    int iret[8];
    cullPoints2 (cnum,ret,maxc,i1,iret);

    for (j=0; j < maxc; j++) {
//      dContactGeom *con = CONTACT(contact,skip*j);
  //    for (i=0; i<3; i++) con->pos[i] = point[iret[j]*3+i] + pa[i];
    //  con->depth = dep[iret[j]];

		btVector3 posInWorld;
		for (i=0; i<3; i++) 
			posInWorld[i] = point[iret[j]*3+i] + pa[i];
		if (code<4) 
	   {
			output.addContactPoint(-normal,posInWorld,-dep[iret[j]]);
		} else
		{
			output.addContactPoint(-normal,posInWorld-normal*dep[iret[j]],-dep[iret[j]]);
		}
    }
    cnum = maxc;
  }

  *return_code = code;
  return cnum;
}