2-manifold bug - repro included

Hi -

I was making a complex polyhedron, and keep encountering the "UI-WARNING:
Object may not be a valid 2-manifold and may need repair!" message upon
rendering. I was pretty sure that my polyhedron was valid, so I kept
simplifying and simplifying until I got to a trivial repro case. See the
below program. The fascinating thing (to me) is that when I just include the
polyhedron, it renders without error. But when I also add the (non-touching)
red cube off the positive x side, then the render step shows an error.

I cannot help but think this is a bug somewhere? I can't see why two
non-intersecting objects would render individually without errors, but when
both are in the same file, there would be an error. Just in case it's
relevant, I'm on OpenSCAD version 2019.05 running on OSX 10.14.6.

color(c = "green") {
        points = [
                [0, 0, 1],    // Lower left front #1
                [0, 20, 0],   // Lower left back
                [10, 20, 0],  // Lower right back
                [10, 0, 1],   // Lower right front #1
                [0, 0, 30],   // Upper left front
                [0, 20, 30],  // Upper left back
                [10, 20, 30], // Upper right back
                [10, 0, 30],  // Upper right front
                [0, 1, 0],    // Lower left front #2
                [10, 1, 0]    // Lower right front #2
        ];
        faces = [
                [9, 2, 1, 8], // Bottom
                [4, 5, 6, 7], // Top
                [2, 3, 7, 6], // Right top
                [3, 2, 9],    // Right bottom
                [5, 4, 0, 1], // Left top
                [1, 8, 0],    // Left bottom
                [4, 7, 3, 0], // Front
                [6, 5, 1, 2], // Back
                [0, 3, 9, 8]  // Angled front-bottom
        ];
        polyhedron(points=points, faces=faces);
}

translate(v = [15, 0, 0]) {
        color(c = "red") {
                cube(size = [10, 20, 30]);
        }
}




--
Sent from: http://forum.openscad.org/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

The keyword is winding order, this face is looking
in the wrong direction (which you can see as pink-ish
color when using F12):

[1, 8, 0],    // Left bottom

If you flip it, e.g. by using:

[1, 0, 8],    // Left bottom

the union works fine.

ciao,
   Torsten.

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

That was the hardest concept for me to 'grok' in GL classes. Still not
sure I got it completely.

On 11/27/20 7:05 PM, Torsten Paul wrote:
_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

It looks like some people have identified your problem (faces in wrong
orientation), but not explained this fact:  if you make a bogus polyhedron
by itself in your model it doesn't get processed by CGAL.  A single
polyhedron will always appear valid.  Only when you add a second object and
force a union computation, for example, does your polyhedron get checked for
validity.   Call it a quirk rather than a bug.  

Depending on what you're trying to do you might find the VNF (vertices 'n
faces) code in the BOSL2 library helpful for constructing your polyhedron.
It can make this task easier to get right, with everything valid.  

https://github.com/revarbat/BOSL2/wiki/vnf.scad


ssuchter wrote
> Discuss@.openscad

> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org





--
Sent from: http://forum.openscad.org/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

Thanks everyone for the help. I'm annoyed that I got that face backwards - I
thought I double checked everything. I understand the quirk about the union
bit now.



--
Sent from: http://forum.openscad.org/

_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org