cut with solid from parametricCurve

[asprionj]

(I'm on 2.0RC1.) Trying to cut a body that is constructed by revolving a parametricCurve from another body. For illustration, a minimal example which is working as expected, but using a circle instead of said parametricCurve:

s = cq.Workplane('XY').circle(50).extrude(20) # the base body
to_cut = cq.Workplane('XY').transformed(offset=[35,0,0], rotate=[90,0,0]).circle(14).revolve(315, [-35,0,0], [-35,1,0]) # this will be cut away
s = s.cut(to_cut) # cut it...

The result looks fine (I cannot fully close the torus because of #259):

Now, if I use a parametricCurve, e.g. to have an ellipse instead of a circle (or any other complex geometry):

from math import sin, cos, pi

def my_ellipse(t, a, b):
    return (a*sin(2*pi*t), b*cos(2*pi*t))

to_cut = cq.Workplane('XY').transformed(offset=[35,0,0], rotate=[90,0,0]).parametricCurve(lambda t: my_ellipse(t,14,7)).revolve(315, [-35,0,0], [-35,1,0])

it fails with "RuntimeError: Standard_Failure Courbes non jointives". (Also when using intersect instead of cut.)

Funny thing is, if I use union instead of cut, it does work as expected:

[adam-urbanczyk]

Confirmed, looks like a cad kernel issue. Same workaround as in #259 :

from math import sin, cos, pi

def my_ellipse(t, a, b):
    return (a*sin(2*pi*t), b*cos(2*pi*t))

to_cut = cq.Workplane('XY').transformed(offset=[35,0,0], rotate=[90,0,0]).parametricCurve(lambda t: my_ellipse(t,14,7)).revolve(360, [-35,0,0], [-35,1,0],clean=False)

s = cq.Workplane('XY').circle(50).extrude(20)
s = s.cut(to_cut,clean=False)

show_object(s)

[lorenzncode]

Fixed in occt7.6 branch.

from math import sin, cos, pi

def my_ellipse(t, a, b):
    return (a*sin(2*pi*t), b*cos(2*pi*t))

s = cq.Workplane('XY').circle(50).extrude(20) # the base body
to_cut = cq.Workplane('XY').transformed(offset=[35,0,0], rotate=[90,0,0]).parametricCurve(lambda t: my_ellipse(t,14,7)).revolve(315, [-35,0,0], [-35,1,0])
s = s.cut(to_cut)

print(s.val().isValid())

occt7.6:

current master (ocp-7.5.3):

[adam-urbanczyk]

Thanks @lorenzncode for checking - I'm adding this issue to the PR.