Memory leak during updating position of tetrapoly elements

[MikeZhao]

There exists severe memory leaking problem when using intermediate variables to update positions of tetraPoly in a loop:

running the code, the memory will keep leaking:
#
a = [[0,0,0],[0,0,1],[0,1,0],[1,0,0]]
O.bodies.append(yade.utils.tetraPoly(a))
for i in range(1000000):
    b = [[0,0,0],[0,0,1],[0,1,0],[i+2,0,0]]
    O.bodies[0].shape.setVertices(b)
    O.bodies[0].state.ori = O.bodies[0].shape.GetOri()
    O.bodies[0].state.pos = O.bodies[0].shape.GetCentroid()
    O.bodies[0].state.mass = O.bodies[0].mat.density * O.bodies[0].shape.GetVolume()
    O.bodies[0].state.inertia = O.bodies[0].mat.density * O.bodies[0].shape.GetInertia()

[Jan Stránský (honzik)]

setVertices now works without memory leaks passing each vertex as single argument

O.bodies[0].shape.setVertices(b[0],b[1],b[2],b[3])

or

O.bodies[0].shape.setVertices(*b) # *b in Python does internally the above code

currently implemented only for 4 vertices, if needed it is easy to extend to more vertices

Jan

[Jan Stránský (honzik)]

In previous comment, I meant
O.bodies[0].shape.setVertices(a[0],a[1],a[2],a[3])
O.bodies[0].shape.setVertices(a)
to be consistent with original bug report