[Bf-blender-cvs] [d17a521] alembic: Fix for NaN error when using zero-length springs in the mass-spring solver.

[Lukas Tönne]

Commit: d17a52109ee59c2f883f81e9bb821a727f4cba2a
Author: Lukas Tönne
Date:   Thu May 7 16:59:31 2015 +0200
Branches: alembic
https://developer.blender.org/rBd17a52109ee59c2f883f81e9bb821a727f4cba2a

Fix for NaN error when using zero-length springs in the mass-spring
solver.

This happens when using goal springs, which are now defined as linear
springs between actual vertices and virtual goal vertices. The Jacobian
was not accounting for the possibility of zero rest length.

===================================================================

M	source/blender/physics/intern/implicit_blender.c

===================================================================

diff --git a/source/blender/physics/intern/implicit_blender.c b/source/blender/physics/intern/implicit_blender.c
index 47bbc30..418baf3 100644
--- a/source/blender/physics/intern/implicit_blender.c
+++ b/source/blender/physics/intern/implicit_blender.c
@@ -1522,10 +1522,15 @@ BLI_INLINE void dfdx_spring(float to[3][3], const float dir[3], float length, fl
 {
 	// dir is unit length direction, rest is spring's restlength, k is spring constant.
 	//return  ( (I-outerprod(dir, dir))*Min(1.0f, rest/length) - I) * -k;
-	outerproduct(to, dir, dir);
-	sub_m3_m3m3(to, I, to);
+	if (L > ALMOST_ZERO) {
+		outerproduct(to, dir, dir);
+		sub_m3_m3m3(to, I, to);
+		
+		mul_m3_fl(to, (L / length)); 
+	}
+	else
+		zero_m3(to);
 	
-	mul_m3_fl(to, (L/length)); 
 	sub_m3_m3m3(to, to, I);
 	mul_m3_fl(to, k);
 }
@@ -1551,7 +1556,7 @@ BLI_INLINE void dfdv_damp(float to[3][3], const float dir[3], float damping)
 
 BLI_INLINE float fb(float length, float L)
 {
-	float x = length / L;
+	float x = L > ALMOST_ZERO ? length / L : 0.0f;
 	float xx = x * x;
 	float xxx = xx * x;
 	float xxxx = xxx * x;
@@ -1560,7 +1565,7 @@ BLI_INLINE float fb(float length, float L)
 
 BLI_INLINE float fbderiv(float length, float L)
 {
-	float x = length/L;
+	float x = L > ALMOST_ZERO ? length / L : 0.0f;
 	float xx = x * x;
 	float xxx = xx * x;
 	return (-46.164f * xxx + 102.579f * xx - 78.166f * x + 23.116f);