On Wed, May 30, 2018 at 2:07 AM, MichaelAtOz oz.at.michael@gmail.com
wrote:

Okay. Thanks. I did find OpenSCAD.net which seems to be able to load files
from Dropbox or some such, but no customizer support --- anyone know if
that's in the pipeline to be added?

William

According to the OpenSCAD.net github repo, "This project is no longer
actively maintained, and was written against an older version of OpenSCAD
(v 2011.06) which has now been superseded."

On 30 May 2018 at 09:39, William Adams will.adams@frycomm.com wrote:

Darn. Thanks.

Here's a thought: has anyone looked into integrating OpenSCAD into Jupyter
Notebooks?

William

On Wed, May 30, 2018 at 12:46 PM, doug moen doug@moens.org wrote:

And it turns out folks have already discussed this:

https://github.com/jupyter/notebook/issues/659

and it's supposed to be straight-forward:

https://github.com/ipython/ipython/wiki/IPython-kernels-for-other-languages.
If OpenSCAD implement the protocol it should just work out of the box
without us having to do anything. Thanks !

William

On Wed, May 30, 2018 at 1:01 PM, William Adams will.adams@frycomm.com
wrote:

And maybe OpenJSCAD.org will work?

On Wed, May 30, 2018 at 1:23 PM, William Adams will.adams@frycomm.com
wrote:

Implementing the protocol is not straight-forward.

Jupyter is a web application. The user interface runs in a web browser. The
kernel runs in a separate process, and communicates with the front end
using a JSON encoded network protocol.

This sounds kind of like reimplementing a lot of OpenSCAD from scratch. And
yeah, OpenJSCAD might be a better starting point.

On 30 May 2018 at 13:23, William Adams will.adams@frycomm.com wrote:

On 05/30/2018 08:06 PM, doug moen wrote:

Why?

There's even a C++ kernel interface library on github:
https://github.com/QuantStack/xeus

ciao,
Torsten.

To do this, you need to create a clean separation of the OpenSCAD GUI from
the OpenSCAD kernel. Put the kernel in a separate library and formalize the
API that connects the kernel and the GUI. Then create a new user interface
that runs in a web browser and communicates with the kernel over Jupyter
protocol, using Xeus on the kernel side. An intermediate step might be for
the kernel to always render the model to a triangle mesh, then send the
mesh to the client. But rendering can take a long time. To do fast
previewing, you need an equivalent of OpenCSG that runs in a web browser.

That's my understanding of what it would mean to have an OpenSCAD kernel
for Jupyter. Does this sound easy and straightforward? Or do you have a
different idea?

On 30 May 2018 at 14:38, Torsten Paul Torsten.Paul@gmx.de wrote:

My idea was just to have Jupyter Notebook launch OpenSCAD and run it on a
separate file, then read in the JSON file with the presets --- I guess
there's some security aspect which makes that more difficult than I was
thinking?

Oh well, worst case is the user will have to manually run OpenSCAD, then
import the files from that, then Jupyter Notebook can operate on those.

I need to finish the lualatex / metapost integration I've started on first
though.

William

On Wed, May 30, 2018 at 3:17 PM, doug moen doug@moens.org wrote:

On 05/30/2018 09:17 PM, doug moen wrote:

That might be the perfect solution, but I'm pretty sure
it's not needed to get a working solution. Otherwise
Thingiverse Customizer would not exist.

ciao,
Torsten.