Hi Larry,

If your scad 2d is the shape you want to extrude and round then probably
the most succinct solution would be to use  NopSCADlib.

Here is an example to do just that, if I understand your requirement
correctly.
Also, that 2d shape, could be generated in plain openscad, it looks like
a large circle with other small circles differenced around the
circumference.

Best wishes,

Ray

//
// NopSCADlib Copyright Chris Palmer 2018
// nop.head@gmail.com
// hydraraptor.blogspot.com
//
// This file is part of NopSCADlib.
//
// NopSCADlib is free software: you can redistribute it and/or modify it
under the terms of the
// GNU General Public License as published by the Free Software
Foundation, either version 3 of
// the License, or (at your option) any later version.
//
// NopSCADlib is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY;
// without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE.
// See the GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
along with NopSCADlib.
// If not, see https://www.gnu.org/licenses/.
//

include <../utils/core/core.scad>
use <../utils/round.scad>

module shape()
    difference() {
        square(40, center = true);

        square([20, 20], center = true);
    }

module rounds() {
    linear_extrude(eps)
        round(or = 4, ir = 2)
            shape();

    translate([50, 0])
        round_3D(or = 4, ir = 2, chamfer_base = true, $fn = 16)
            linear_extrude(40, center = true)
                shape();
}

rounds();

On 12/10/2026 22:50, larry via Discuss wrote:

On Tue, 2026-10-13 at 13:20 -0700, Jordan Brown wrote:

Yes that's about right.

Yes, I need a flat bottom, but I can definitely manage that. I will be
putting this on a low cylinder, which can eat the rounded bottom

Sounds good! 

I was looking at something that looked like it extruded a star with a
rounded top. That was what I could not do with my path.

Thanks for the pointer!

On 10/14/2026 1:24 AM, larry wrote:

The joint_bot parameter controls the rounding on the bottom.

On Wed, 2026-10-14 at 08:32 -0700, Jordan Brown wrote:

Figured that out from the code you sent, but I now have another
problem.

I get an error when I try to preview or render. It looks like the
points are messed up. As I mentioned, I used ChatGPT to convert from a
projection() to points, and in all likelihood it got that wrong.

I've attached the code I attempted. I don't really understand the error
message.

I don't mind building the path from a large circle, differenced with
small circles, but I have no idea how to make a polygon from that.

Larry

On 10/14/2026 9:59 AM, larry wrote:

I don't get an error previewing or rendering that file, on either
2021.01 or a development snapshot.

What message do you get?

A polygon, in the sense of a 2D shape, is easy:  the result of the
difference is a polygon.

What you can't get from base OpenSCAD is a list of points, as would be
required for a sweep or rounded_prism().

BOSL2 can do 2D differences on its own, so you could do all of the
differencing "in user space" to yield a list of points.

Here's a demonstration, give or take the exact parameters.

include <BOSL2/std.scad>

$fa = 1;
$fs = 1;

bigR = 70;
smallCenter = 71;
smallR = 6;
nSmall = 15;

big = circle(r=bigR);
small = fwd(smallCenter, circle(r=smallR));
smalls = [ for (i = [0:nSmall-1]) zrot(i * 360/nSmall, small) ];
result = difference(big, smalls);
region(result);

Note that a BOSL2 "region" is a list of lists of points.  (The result of
a difference needs to be such a list, because it might be multiple
shapes.)  In this particular case, where we know that there's only one
shape, you could use polygon(result[0]);

On Wed, 2026-10-14 at 10:21 -0700, Jordan Brown wrote:

I'm running OpenSCAD version 2026.08.03, and I get the attached.
 

I'll have a look at this later.

Thanks Jordan.

It looks like you have a big point list and furthermore, it's full of
collinear points.  You can use path_merge_collinear to clean that up.  Most
processing in BOSL2 doesn't like collinear points.  Also your point list
appears long.  The rounded_prism() module is good when you have a polygon
with a few edges, like 10, say.  When you have a more continuous path it's
better to use offset_sweep because rounded_prism will make hundreds of tiny
roundings between every single pair of line segments in your path.  It also
looks like your code has an error that's making it think joint_bot is a
function, but I don't see your code here.

On Wed, Oct 14, 2026 at 15:18 larry via Discuss discuss@lists.openscad.org
wrote:

On 10/14/2026 12:15 PM, larry via Discuss wrote:

The example you gave earlier just called polygon(); it didn't call
rounded_prism().

When I wrap a rounded_prism() around those points, I get that same
error.  (Note that it's an error from BOSL2, not from OpenSCAD per se.)

I understand what it's saying:  that there are duplicate or colinear
points.  I don't immediately understand why it cares, though.

I don't know what the real limitations are, but the documentation for
rounded_prism() says:

You do have a lot of points.  You could try dramatically reducing the
number of points.