translate(path_points[0])
     rotate([0, 180 / $fn, 0]) sphere(ballDiam / 2);

 polyhedron(points = tube_points, faces = concat([cap(facets)],

 translate(path_points[segments - 1])
     sphere(ballDiam / 2);

         }
     }

 };

Actually the rotation is only needed when $fn is a multiple of 4. It lines up without when $fn is a multiple of 2 and not at all if it is odd. On 14 June 2015 at 15:41, nop head <nop.head@gmail.com> wrote: > Actually it was caused by the winding order of one of the end caps being > wrong. This fixes it: - > > function cap(facets, segment = 0, i = 0) = i < facets ? concat(segment ? > facets * segment + i : facets - 1 - i, cap(facets, segment, i + 1)) : []; > > Also the sphere facets can be aligned with the tube by adding a rotation: > > translate(path_points[0]) > rotate([0, 180 / $fn, 0]) sphere(ballDiam / 2); > > > > > On 14 June 2015 at 15:17, nop head <nop.head@gmail.com> wrote: > >> PolySet has nonplanar faces. Attempting alternate construction >> >> ERROR: Alternate construction failed. CGAL error in >> CGAL_Nef_polyhedron3(): CGAL ERROR: assertion violation! Expr: >> e->incident_sface() != SFace_const_handle() File: >> /data/OpenSCAD/libraries-mingw32-master/mxe/usr/i686-w64-mingw32.static/include/CGAL/Nef_S2/SM_const_decorator.h >> Line: 326 >> >> >> >> On 14 June 2015 at 15:07, nop head <nop.head@gmail.com> wrote: >> >>> Unfortunately F6 fails on my system but it works in 2 minutes if I >>> remove the spheres at the end points. >>> >>> Looks like a bug in OpenScad triggered by unioning nearly coincident >>> faces. >>> >>> On 14 June 2015 at 14:54, jon <jon@jonbondy.com> wrote: >>> >>>> Thank you very much. You are very generous with your time. >>>> >>>> Jon >>>> >>>> >>>> On 6/14/2015 8:31 AM, nop head wrote: >>>> >>>> rch = 0.1; >>>> ballDiam = 25.4 * 1/4 + rch; >>>> myScale = 1/40; >>>> hubR = 5; >>>> >>>> function f(a,b,t) = // spiral >>>> [ (a + t * b) * cos(t), >>>> (a + t * b) * sin(t), >>>> 0 >>>> ]; >>>> >>>> $fn = 50; >>>> >>>> function m_translate(v) = [ [1, 0, 0, 0], >>>> [0, 1, 0, 0], >>>> [0, 0, 1, 0], >>>> [v.x, v.y, v.z, 1 ] ]; >>>> >>>> function m_rotate(v) = [ [1, 0, 0, 0], >>>> [0, cos(v.x), sin(v.x), 0], >>>> [0, -sin(v.x), cos(v.x), 0], >>>> [0, 0, 0, 1] ] >>>> * [ [ cos(v.y), 0, -sin(v.y), 0], >>>> [0, 1, 0, 0], >>>> [ sin(v.y), 0, cos(v.y), 0], >>>> [0, 0, 0, 1] ] >>>> * [ [ cos(v.z), sin(v.z), 0, 0], >>>> [-sin(v.z), cos(v.z), 0, 0], >>>> [ 0, 0, 1, 0], >>>> [ 0, 0, 0, 1] ]; >>>> >>>> function vec3(v) = [v.x, v.y, v.z]; >>>> function transform(v, m) = vec3([v.x, v.y, v.z, 1] * m); >>>> >>>> function orientate(position, normal) = m_rotate([0, >>>> atan2(sqrt(pow(normal[0], 2) + pow(normal[1], 2)), normal[2]), 0]) >>>> * m_rotate([0, 0, atan2(normal[1], normal[0])]) >>>> * m_translate(position); >>>> >>>> function circle_points(r = 1, a = 0) = a < 360 ? concat([[r * sin(a), r >>>> * cos(a), 0]], circle_points(r, a + 360 / $fn)) : [] ; >>>> >>>> function path_points(a, b, step, t = 0) = t <= 720 ? concat([f(a,b,t)], >>>> path_points(a, b, step, t + step)) : [] ; >>>> >>>> function transform_points(list, matrix, i = 0) = i < len(list) ? >>>> concat([ transform(list[i], matrix) ], transform_points(list, matrix, i + >>>> 1)) : []; >>>> >>>> function tube_points(profile, path, i = 0) = i < len(path) - 1 >>>> ? concat(transform_points(profile, >>>> orientate(path[i], (path[i + 1] - path[i])/ 2)), tube_points(profile, path, >>>> i + 1)) >>>> : transform_points(profile, >>>> orientate(path[i], (path[i] - path[i - 1])/ 2)) ; >>>> >>>> function tube_faces(segs, facets, s, i = 0) = i < facets ? concat([[s * >>>> facets + i, >>>> s * >>>> facets + (i + 1) % facets, >>>> ((s >>>> + 1) % segs) * facets + (i + 1) % facets, >>>> ((s >>>> + 1) % segs) * facets + i]], >>>> >>>> tube_faces(segs, facets, s, i + 1)) : []; >>>> >>>> function path_faces(segs, facets, i = 0) = i < segs - 1 ? >>>> concat(tube_faces(segs, facets, i), path_faces(segs, facets, i + 1)) : []; >>>> >>>> function cap(facets, segment = 0, i = 0) = i < facets ? concat(i + >>>> facets * segment, cap(facets, segment, i + 1)) : []; >>>> >>>> circle_points = circle_points(ballDiam / 2); >>>> facets = len(circle_points); >>>> >>>> path_points = path_points(hubR, myScale, 1); >>>> segments = len(path_points); >>>> >>>> tube_points = tube_points(circle_points, path_points); >>>> path_faces = path_faces(segments, facets); >>>> >>>> union() { >>>> translate(path_points[0]) >>>> sphere(ballDiam / 2); >>>> >>>> polyhedron(points = tube_points, faces = concat([cap(facets)], >>>> path_faces, [cap(facets, segments - 1)])); >>>> >>>> translate(path_points[segments - 1]) >>>> sphere(ballDiam / 2); >>>> } >>>> >>>> >>>> On 14 June 2015 at 12:41, jon <jon@jonbondy.com> wrote: >>>> >>>>> I found and modified kit wallace's knot code (which is based on some >>>>> of your code). Once I understood that the function at the top controlled >>>>> the entire object, it was indeed simple. That is, the two of you made it >>>>> simple for me! Not sure how to eliminate the requirement that the tube be >>>>> closed, though. >>>>> >>>>> On 6/13/2015 8:55 PM, nop head wrote: >>>>> >>>>> $fn = 50; >>>>> for (d = [0:1:720]) hull() >>>>> >>>>> I think it will just take a very long time or consume all of the >>>>> memory because CGAL hates the massive implicit union of objects with lots >>>>> of vertices. It simply doesn't scale to high detail models. If you can >>>>> construct it with functions that generate lists to feed into polyhedron it >>>>> will be much faster. >>>>> >>>>> See the work kitwallace did with knots. A spiral is not too different. >>>>> >>>>> On 14 June 2015 at 01:35, jon <jon@jonbondy.com> wrote: >>>>> >>>>>> I can get this to work with F5, but not with F6: it just hangs for >>>>>> hours, running full CPU. Any thoughts? >>>>>> >>>>>> >>>>>> rch = 0.1; >>>>>> ballDiam = 25.4 * 1/4 + rch; >>>>>> $fn = 50; >>>>>> >>>>>> module spiral() { >>>>>> myScale = 1/40; >>>>>> hubR = 5; >>>>>> for (d = [0:1:720]) { >>>>>> hull(){ >>>>>> rotate([0, 0, d]) >>>>>> translate([hubR + d * myScale, 0, 0]) >>>>>> sphere(d = ballDiam); >>>>>> rotate([0, 0, d]) >>>>>> translate([hubR + (d+1) * myScale, 0, 0]) >>>>>> sphere(d = ballDiam); >>>>>> >>>>>> } >>>>>> } >>>>>> >>>>>> }; >>>>>> >>>>>> difference() { >>>>>> translate([0, 0, -ballDiam * 0.7]) >>>>>> cylinder(h=ballDiam, r=28); >>>>>> spiral(); >>>>>> } >>>>>> >>>>>> _______________________________________________ >>>>>> OpenSCAD mailing list >>>>>> Discuss@lists.openscad.org >>>>>> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >>>>>> >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> OpenSCAD mailing listDiscuss@lists.openscad.orghttp://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >>>>> >>>>> >>>>> >>>>> No virus found in this message. >>>>> Checked by AVG - www.avg.com >>>>> Version: 2015.0.6030 / Virus Database: 4360/10009 - Release Date: >>>>> 06/13/15 >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> OpenSCAD mailing list >>>>> Discuss@lists.openscad.org >>>>> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >>>>> >>>>> >>>> >>>> >>>> _______________________________________________ >>>> OpenSCAD mailing listDiscuss@lists.openscad.orghttp://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >>>> >>>> >>>> >>>> No virus found in this message. >>>> Checked by AVG - www.avg.com >>>> Version: 2015.0.6030 / Virus Database: 4360/10012 - Release Date: >>>> 06/14/15 >>>> >>>> >>>> >>>> _______________________________________________ >>>> OpenSCAD mailing list >>>> Discuss@lists.openscad.org >>>> http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >>>> >>>> >>> >> >

I'm hitting this as well. I wrote a module to do "shift_extrude" and it works nicely with F5 preview. For F6 rendering, if I only have one shifted extrusion, it's fine, but if I add a second, I get: ERROR: CGAL error in CGAL_Nef_polyhedron3(): CGAL ERROR: assertion violation! Expr: e->incident_sface() != SFace_const_handle() File: /opt/mxe/usr/x86_64-w64-mingw32.static/include/CGAL/Nef_S2/SM_const_decorator.h Line: 326 I extracted the minimum code to reproduce the error: bug_report.scad <http://forum.openscad.org/file/n20483/bug_report.scad> Oddly enough, it doesn't seem to be a matter of the number of slices. I tried changing the number of slices to 10 to reduce the number of vertices and it still fails. Even with the separation between the two objects. -- View this message in context: http://forum.openscad.org/F5-works-but-not-F6-tp12875p20483.html Sent from the OpenSCAD mailing list archive at Nabble.com.

On 02/19/2017 04:03 AM, bdl wrote: > I extracted the minimum code to reproduce the error: > bug_report.scad <http://forum.openscad.org/file/n20483/bug_report.scad> > The winding order of some faces is wrong. See https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/FAQ#Why_is_my_model_showing_up_with_F5_but_not_F6.3F ciao, Torsten.

Yeah, I just figured that out while playing around. My regular_polygon winds counterclockwise in the plane, to the top and bottom faces are wrong if height > 0 and all the other faces are wrong if height < 0. I fixed it and now things seem to be working. What's odd is I've had errors in face winding before and never gotten that error... usually things just look funny... but maybe I was only using F5 in those cases and just didn't notice the problem here. -- View this message in context: http://forum.openscad.org/F5-works-but-not-F6-tp12875p20485.html Sent from the OpenSCAD mailing list archive at Nabble.com.

well done nophead. This is a version using my sweep(): > use > <Naca_sweep.scad> > $fn=50; > spiral(a1 = 120, a2 = 960, r=22, rc=10); > > module spiral(a1=180, a2=720, r=25, rc=10) > { > sweep(gendat(a1, a2, r, rc)); > translate(spiral_(a1, r)) sphere(rc); > translate(spiral_(a2, r)) sphere(rc); > } > > function gendat(a1=180, a2=720, r=25, rc=10) = let(N=$fn?$fn:360/$fa, > M=ceil((a2-a1)/360*N), st = (a2-a1)/M) > [for(i=[0:M]) let(a=i*st+a1) traject(circle_(rc), direction_(a), > spiral_(a, r))]; // compose all trajectory data > > function circle_(r=10, n=undef) = let(N=n?n:$fn?$fn:360/$fa, st = 360/N) > [for(i=[0:N-1]) [r*cos(i*st), 0, r*sin(i*st)]]; // 2D shape > > function spiral_(a, r) = r*a/360*[sin(a), cos(a)]; // archimedes spiral > > function direction_(a, r=10) = let(p1 = spiral_(a-.5, r), p2 = > spiral_(a+.5, r)) (p2-p1)/norm(p2-p1); // direction vector needed for > rotation > > function traject(L, v, t) = let(rot = [[v[1], -v[0], 0],[-v[0], v[1], 0], > [0, 0, 1]]) > [for(i=L) rot*i+[t[0], t[1], 0]]; // rotate + translate to place circle > into trajectory -- View this message in context: http://forum.openscad.org/F5-works-but-not-F6-tp12875p20493.html Sent from the OpenSCAD mailing list archive at Nabble.com.

Nice. A bit simpler than mine! On 19 February 2017 at 18:53, Parkinbot <rudolf@parkinbot.com> wrote: > well done nophead. This is a version using my sweep(): > > > > use > > <Naca_sweep.scad> > > $fn=50; > > spiral(a1 = 120, a2 = 960, r=22, rc=10); > > > > module spiral(a1=180, a2=720, r=25, rc=10) > > { > > sweep(gendat(a1, a2, r, rc)); > > translate(spiral_(a1, r)) sphere(rc); > > translate(spiral_(a2, r)) sphere(rc); > > } > > > > function gendat(a1=180, a2=720, r=25, rc=10) = let(N=$fn?$fn:360/$fa, > > M=ceil((a2-a1)/360*N), st = (a2-a1)/M) > > [for(i=[0:M]) let(a=i*st+a1) traject(circle_(rc), direction_(a), > > spiral_(a, r))]; // compose all trajectory data > > > > function circle_(r=10, n=undef) = let(N=n?n:$fn?$fn:360/$fa, st = 360/N) > > [for(i=[0:N-1]) [r*cos(i*st), 0, r*sin(i*st)]]; // 2D shape > > > > function spiral_(a, r) = r*a/360*[sin(a), cos(a)]; // archimedes spiral > > > > function direction_(a, r=10) = let(p1 = spiral_(a-.5, r), p2 = > > spiral_(a+.5, r)) (p2-p1)/norm(p2-p1); // direction vector needed for > > rotation > > > > function traject(L, v, t) = let(rot = [[v[1], -v[0], 0],[-v[0], v[1], 0], > > [0, 0, 1]]) > > [for(i=L) rot*i+[t[0], t[1], 0]]; // rotate + translate to place circle > > into trajectory > > > > > > -- > View this message in context: http://forum.openscad.org/F5- > works-but-not-F6-tp12875p20493.html > Sent from the OpenSCAD mailing list archive at Nabble.com. > > _______________________________________________ > OpenSCAD mailing list > Discuss@lists.openscad.org > http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org >

With a bit more code, the spherical caps may be include in the polyhedron definition. The following code is a self contained specialized sweep for circular sections with spherical caps. The path can be easily changed from the spiral to something else. F6 takes virtually nothing and don't crash when boolean operated if the polyhedron has no self-intersections. <code> R_ini = 20; R_end = 100; ang_ini = 0; ang_end = 800; long_n = 200; sect_n = 32; diam = 20; spiral_path = spiral(R_ini, R_end, ang_ini, ang_end, long_n); spiral_tube = tube(spiral_path, diam/2, sect_n); difference(){ mesh2poly(spiral_tube); // uncomment to check F6 // translate([(R_ini+R_end)/5, (R_ini+R_end)/5, 0]) cube(3*diam); } // build a mesh for the whole tube with spherical caps function tube(path, r, n) = let( m= len(path), tube_ = [for(i=[0:m-1] ) // the tube itself let( pt = path[i], // tangent at path[i] tg = unit(path[min(i+1,m-1)]-path[max(i-1,0)]), c = tg.y, s = -tg.x, cir = circle_(r, n)*[[c,s,0],[-s,c,0],[0,0,1]]) [for(p=cir) p+pt] ], tgi = path[0]-path[1], tgf = path[m-1]-path[m-2], capi = spherical_cap(tube_[0], r*unit(tgi), path[0], true, floor(n/4)), capf = spherical_cap(tube_[m-1], r*unit(tgf), path[m-1], false, floor(n/4)) ) concat( capi, tube_ , capf ); // a half sphere to close the tube function spherical_cap(cir, tg, ct, start, n) = let( m = len(cir), cir0 = [for(p=cir) p-ct ], rang = start? [n:-1:0] : [0:n] ) [for(i=rang) let(c=cos(90*i/n)*cir0, s=sin(90*i/n)) [for(p=c) p+s*tg+ct ] ]; // a spiraled path function spiral(r1, r2, a1, a2, m) = let( da = (a2-a1)/m ) [for(i=[0:m], ai = a1 + i*da, ri = r1 + (r2-r1)*i/m ) ri*[cos(ai), sin(ai),0*cos(ai/3)] ]; function circle_(r,n) = [for(i=[0:n]) r*[cos(360*i/n),0,sin(360*i/n)] ] ; // draw a mesh (matrix of points) as a polyhedron module mesh2poly(rmesh) { n = len(rmesh) ; m = len(rmesh[0]) ; vertices = [ for(i=[n-1: -1: 0]) for(pt=rmesh[i]) pt ]; tris = concat( [ for(i=[0:n-2],j=[0:m-2]) [ i*m+j, (i+1)*m+j, (i+1)*m+j+1 ] ] , [ for(i=[0:n-2],j=[0:m-2]) [ i*m+j, (i+1)*m+j+1, i*m+j+1 ] ] ); polyhedron(vertices, tris); } function unit(v) = v/norm(v); </code> -- View this message in context: http://forum.openscad.org/F5-works-but-not-F6-tp12875p20523.html Sent from the OpenSCAD mailing list archive at Nabble.com.