MM
Michael Marx (spin)
Sat, Oct 10, 2026 2:56 PM
Roel,
Oh well, oops. That also went to the Mailing-list.
So there is your announcement.
I would prefer, that some other contact is established for any feedback regarding 'SCADMaxx'.
No doubt there may be some initial discussion, but details do not belong here.
Michael
OpenSCAD Admin.
-----Original Message-----
From: Michael Marx (spin) [mailto:michael.marx@spin.net.au]
Sent: Sunday, October 11, 2026 1:43 AM
To: 'Roel Vanhout'; 'OpenSCAD general discussion'
Subject: [Admin review] RE: An experimental OpenSCAD build that integrates a few feature
PRs and some new features
Hi Roel, (BCC's a moderated post for your opinion)
I'm one of the OpenSCAD Admin team.
New subscribers to the OpenSCAD Mailing-list are moderated by default.
Hence your first post, below, is up for review before going to the list.
As I read it, you have a 'non-fork', 'experimental build' SCADMaxx, 'incompatible with
OpenSCAD'. Which seem to depart considerably from OpenSCAD.
And "this has been done by agents. Maybe it's all AI slop and a useless waste of time"
And you'd "love to hear from other who've looked at this what solutions you came up with
for the obvious first wave of problems to solve." ..."Feedback welcome!"
So it is NOT OpenSCAD. You seem to be suggesting that the OpenSCAD Mailing-list be
infrastructure for your invention to develop.
I don't see how that helps current OpenSCAD users.
It seems to be categorically much like PythonSCAD. (https://www.pythonscad.org)
Which has it's own web site and support.
I will await comment from the BCC'd other members of the team.
Until then, I will not be approving your post to go to the Mailing-list.
I may be convinced about an announcement, but I don't think your 'problems to solve'
belong on the OpenSCAD Mailing-list.
Michael
OpenSCAD Admin.
---=====================
-----Original Message-----
From: Roel Vanhout [mailto:roel.vanhout@gmail.com]
Sent: Saturday, October 10, 2026 9:13 PM
To: OpenSCAD general discussion
Subject: An experimental OpenSCAD build that integrates a few feature PRs and some new
features
Hello everyone,
I have made an experimental build of OpenSCAD that integrates
openscad-master, some PRs in the queue, as well as some
new features. I'm calling this build SCADMaxx to make sure it doesn't
get confused with OpenSCAD proper, as it does add
language features that make it (forward) incompatible with OpenSCAD.
This is not a fork, it's an experimental build that
I primarily made for myself, but that I hope can also be useful to
other, both for modeling as well as being a testing
ground for new features and iterating on ideas that people have on
what OpenSCAD should do or become.
Some of the features:
- Agentic coding support. HTTP endpoints that expose source, camera views etc.
- Agent annotations. Annotate 3D models live on screen for access by
agents, so you don't have to take screenshots or
explain your issues. Just draw what is wrong or how you want things
to be, and let your agents access that.
- A UI refresh - clean up typography, spacing, coloring, consistent
icons that work across light and dark themes, ...
- Navigation cube in the 3D view, like many other 3D modeling software
packages have. Never again getting lost on where
you're looking from or what you're looking at!
- Live cross sections. Look inside your model by setting up cross
section planes.
- Contextual insertion. No more looking around for module names in
BOSL or BOLTS. Click in the 3D view, shift-A, search
for the part you want to insert (a nut, a bolt, ...), parameterize
it, SCADMaxx inserts the correct code at the
correct location (well, sometimes) in your souce code. This is not
visual modeling! It's assisted code editing.
- A Vim mode for the text editor. I don't use the native text editor,
but I'm hoping that with native Vim mode I will
again :)
- A launcher - if you have many builds of OpenSCAD on your machine,
you'll know the pain of wondering 'what build am I
looking at' or 'I want to open this file with this build, but where
is it'. The launcher lets you manage builds and
associate .scad and .scadmaxx file extensions with it.
- Customizable keyboard shortcuts (PR #6878).
- Background preview and rendering. Rendering doesn't happen on the UI
thread any more, so it can be cancelled and the
UI stays responsive. Never again 'oh no I accidentally opened this
model that takes 10 seconds to render'.
- Object literals (PR #6081).
- Dynamic customizer annotations (PR #6642, but with the object
literals syntax). (Amongst other things,) the Customizer
can now change what it shows at runtime, so when you select one
property of an object, you'll only see other
properties that are relevant in that case.
- Parts, Assemblies and Components. Language-level annotations that
let you build and integrate at a much higer level
than you could before.
A full list of features with screenshots is on
https://github.com/roel-v/SCADMaxx . There's a Windows prebuild binary
inthe releases (I mean there's source for all of this too, of course).
I haven't had time for a Linux build, and I don't
have aMac for building it for that.
Obviously, most of this has been done by agents. Maybe it's all AI
slop and a useless waste of time and resources. Time
will tell. For now, it's been useful for me at least, and that was
the reason I set out to do it. But what I want to
say is, if you're going to look at code and judge it by 2024 standards
and aren't used to agentic coding, you'll be
sorely disappointed, very angry and/or physically ill. Just a friendly
warning up front :)
A future thing I've been experimenting with is an OpenCASCADE backend,
in addition to the CGAL and Manifold ones. That's
not in this build, it's too early, but my early investigation is
(somewhat) promising. So if you have ideas about this,
do let me know! I'd love to hear from other who've looked at this what
solutions you came up with for the obvious first
wave of problems to solve.
Feedback welcome!
cheers,
Roel
Roel,
Oh well, oops. That also went to the Mailing-list.
So there is your announcement.
I would prefer, that some other contact is established for any feedback regarding 'SCADMaxx'.
No doubt there may be some initial discussion, but details do not belong here.
Michael
OpenSCAD Admin.
> -----Original Message-----
> From: Michael Marx (spin) [mailto:michael.marx@spin.net.au]
> Sent: Sunday, October 11, 2026 1:43 AM
> To: 'Roel Vanhout'; 'OpenSCAD general discussion'
> Subject: [Admin review] RE: An experimental OpenSCAD build that integrates a few feature
> PRs and some new features
>
> Hi Roel, (BCC's a moderated post for your opinion)
>
> I'm one of the OpenSCAD Admin team.
>
> New subscribers to the OpenSCAD Mailing-list are moderated by default.
> Hence your first post, below, is up for review before going to the list.
>
> As I read it, you have a 'non-fork', 'experimental build' SCADMaxx, 'incompatible with
> OpenSCAD'. Which seem to depart considerably from OpenSCAD.
>
> And "this has been done by agents. Maybe it's all AI slop and a useless waste of time"
>
> And you'd "love to hear from other who've looked at this what solutions you came up with
> for the obvious first wave of problems to solve." ..."Feedback welcome!"
>
> So it is NOT OpenSCAD. You seem to be suggesting that the OpenSCAD Mailing-list be
> infrastructure for your invention to develop.
>
> I don't see how that helps current OpenSCAD users.
> It seems to be categorically much like PythonSCAD. (https://www.pythonscad.org)
> Which has it's own web site and support.
>
> I will await comment from the BCC'd other members of the team.
> Until then, I will not be approving your post to go to the Mailing-list.
>
> I may be convinced about an announcement, but I don't think your 'problems to solve'
> belong on the OpenSCAD Mailing-list.
>
>
> Michael
> OpenSCAD Admin.
>
> =======================================================================================
> > -----Original Message-----
> > From: Roel Vanhout [mailto:roel.vanhout@gmail.com]
> > Sent: Saturday, October 10, 2026 9:13 PM
> > To: OpenSCAD general discussion
> > Subject: An experimental OpenSCAD build that integrates a few feature PRs and some new
> > features
> >
> > Hello everyone,
> >
> > I have made an experimental build of OpenSCAD that integrates
> > openscad-master, some PRs in the queue, as well as some
> > new features. I'm calling this build SCADMaxx to make sure it doesn't
> > get confused with OpenSCAD proper, as it does add
> > language features that make it (forward) incompatible with OpenSCAD.
> > This is not a fork, it's an experimental build that
> > I primarily made for myself, but that I hope can also be useful to
> > other, both for modeling as well as being a testing
> > ground for new features and iterating on ideas that people have on
> > what OpenSCAD should do or become.
> >
> > Some of the features:
> >
> > - Agentic coding support. HTTP endpoints that expose source, camera views etc.
> > - Agent annotations. Annotate 3D models live on screen for access by
> > agents, so you don't have to take screenshots or
> > explain your issues. Just draw what is wrong or how you want things
> > to be, and let your agents access that.
> > - A UI refresh - clean up typography, spacing, coloring, consistent
> > icons that work across light and dark themes, ...
> > - Navigation cube in the 3D view, like many other 3D modeling software
> > packages have. Never again getting lost on where
> > you're looking from or what you're looking at!
> > - Live cross sections. Look inside your model by setting up cross
> > section planes.
> > - Contextual insertion. No more looking around for module names in
> > BOSL or BOLTS. Click in the 3D view, shift-A, search
> > for the part you want to insert (a nut, a bolt, ...), parameterize
> > it, SCADMaxx inserts the correct code at the
> > correct location (well, sometimes) in your souce code. This is not
> > visual modeling! It's assisted code editing.
> > - A Vim mode for the text editor. I don't use the native text editor,
> > but I'm hoping that with native Vim mode I will
> > again :)
> > - A launcher - if you have many builds of OpenSCAD on your machine,
> > you'll know the pain of wondering 'what build am I
> > looking at' or 'I want to open this file with this build, but where
> > is it'. The launcher lets you manage builds and
> > associate .scad and .scadmaxx file extensions with it.
> > - Customizable keyboard shortcuts (PR #6878).
> > - Background preview and rendering. Rendering doesn't happen on the UI
> > thread any more, so it can be cancelled and the
> > UI stays responsive. Never again 'oh no I accidentally opened this
> > model that takes 10 seconds to render'.
> > - Object literals (PR #6081).
> > - Dynamic customizer annotations (PR #6642, but with the object
> > literals syntax). (Amongst other things,) the Customizer
> > can now change what it shows at runtime, so when you select one
> > property of an object, you'll only see other
> > properties that are relevant in that case.
> > - Parts, Assemblies and Components. Language-level annotations that
> > let you build and integrate at a much higer level
> > than you could before.
> >
> > A full list of features with screenshots is on
> > https://github.com/roel-v/SCADMaxx . There's a Windows prebuild binary
> > inthe releases (I mean there's source for all of this too, of course).
> > I haven't had time for a Linux build, and I don't
> > have aMac for building it for that.
> >
> > Obviously, most of this has been done by agents. Maybe it's all AI
> > slop and a useless waste of time and resources. Time
> > will tell. For now, it's been useful for me at least, and that was
> > the reason I set out to do it. But what I want to
> > say is, if you're going to look at code and judge it by 2024 standards
> > and aren't used to agentic coding, you'll be
> > sorely disappointed, very angry and/or physically ill. Just a friendly
> > warning up front :)
> >
> > A future thing I've been experimenting with is an OpenCASCADE backend,
> > in addition to the CGAL and Manifold ones. That's
> > not in this build, it's too early, but my early investigation is
> > (somewhat) promising. So if you have ideas about this,
> > do let me know! I'd love to hear from other who've looked at this what
> > solutions you came up with for the obvious first
> > wave of problems to solve.
> >
> > Feedback welcome!
> >
> > cheers,
> >
> > Roel