Even if one vertex is collinear, the other one could be reflex;
triangle merging code was screwing up. [git-p4: depot-paths = "//depot/solvespace/": change = 1744]solver
parent
c58dbf1676
commit
d750344653
4
mesh.cpp
4
mesh.cpp
|
@ -120,9 +120,9 @@ void SMesh::Simplify(int start) {
|
|||
bDot /= min(ab.Magnitude(), bc.Magnitude());
|
||||
dDot /= min(cd.Magnitude(), de.Magnitude());
|
||||
|
||||
if(fabs(bDot) < LENGTH_EPS) {
|
||||
if(fabs(bDot) < LENGTH_EPS && dDot > 0) {
|
||||
conv[j] = c;
|
||||
} else if(fabs(dDot) < LENGTH_EPS) {
|
||||
} else if(fabs(dDot) < LENGTH_EPS && bDot > 0) {
|
||||
conv[WRAP((j+1), convc)] = c;
|
||||
} else if(bDot > 0 && dDot > 0) {
|
||||
// conv[j] is unchanged, conv[j+1] goes to [j+2]
|
||||
|
|
Loading…
Reference in New Issue