discuss@lists.openscad.org

OpenSCAD general discussion Mailing-list

View all threads

BelfrySCAD

JB
Jordan Brown
Fri, Oct 23, 2026 10:22 PM

On 10/23/2026 3:10 PM, Todd Allen wrote:

I like the way dragging stuff works in Freecad's sketcher.  It's not
perfect and it has taken a long time to get as good as it is although
the pace of improvement has accelerated, perhaps from AI assisted
coding.  The sketcher is based on defining constraints and a
constraint solver engine that determines what can move and how things
can move when a point is dragged or a constraint such as a distance
between points or an angle or radius is modified.  I think a
constraint system could be a great addition to SCAD either with or
without a GUI. 

There are such systems, but they are very different in style from OpenSCAD.

I don't know how they handle the cube(a+b) case, though.  My guess is
that they apparently-randomly pick one or the other variable to change,
as long as all of the rest of the constraints are still satisfied.

On 10/23/2026 3:10 PM, Todd Allen wrote: > I like the way dragging stuff works in Freecad's sketcher.  It's not > perfect and it has taken a long time to get as good as it is although > the pace of improvement has accelerated, perhaps from AI assisted > coding.  The sketcher is based on defining constraints and a > constraint solver engine that determines what can move and how things > can move when a point is dragged or a constraint such as a distance > between points or an angle or radius is modified.  I think a > constraint system could be a great addition to SCAD either with or > without a GUI.  There are such systems, but they are very different in style from OpenSCAD. I don't know how they handle the cube(a+b) case, though.  My guess is that they apparently-randomly pick one or the other variable to change, as long as all of the rest of the constraints are still satisfied.
TA
Todd Allen
Fri, Oct 23, 2026 10:48 PM

Things only move or change as needed to satisfy constraints.  If you don't
like how something responds you add or change constraints. The goal is to
reach a fully constrained design with a single solution.  For a cube each
of x, y, z, width, height and depth could be constrained by assigning a
value.  Or any of the corner points might be constrained by assigning their
positions or binding them to something else.  As you go the system gives
feedback on what elements are fully constrained, partially constrained,
unconstrained or over constrained.

On Fri, Oct 23, 2026, 5:23 PM Jordan Brown openscad@jordan.maileater.net
wrote:

On 10/23/2026 3:10 PM, Todd Allen wrote:

I like the way dragging stuff works in Freecad's sketcher.  It's not
perfect and it has taken a long time to get as good as it is although the
pace of improvement has accelerated, perhaps from AI assisted coding.  The
sketcher is based on defining constraints and a constraint solver engine
that determines what can move and how things can move when a point is
dragged or a constraint such as a distance between points or an angle or
radius is modified.  I think a constraint system could be a great addition
to SCAD either with or without a GUI.

There are such systems, but they are very different in style from OpenSCAD.

I don't know how they handle the cube(a+b) case, though.  My guess is that
they apparently-randomly pick one or the other variable to change, as long
as all of the rest of the constraints are still satisfied.

Things only move or change as needed to satisfy constraints. If you don't like how something responds you add or change constraints. The goal is to reach a fully constrained design with a single solution. For a cube each of x, y, z, width, height and depth could be constrained by assigning a value. Or any of the corner points might be constrained by assigning their positions or binding them to something else. As you go the system gives feedback on what elements are fully constrained, partially constrained, unconstrained or over constrained. On Fri, Oct 23, 2026, 5:23 PM Jordan Brown <openscad@jordan.maileater.net> wrote: > On 10/23/2026 3:10 PM, Todd Allen wrote: > > I like the way dragging stuff works in Freecad's sketcher. It's not > perfect and it has taken a long time to get as good as it is although the > pace of improvement has accelerated, perhaps from AI assisted coding. The > sketcher is based on defining constraints and a constraint solver engine > that determines what can move and how things can move when a point is > dragged or a constraint such as a distance between points or an angle or > radius is modified. I think a constraint system could be a great addition > to SCAD either with or without a GUI. > > > There are such systems, but they are very different in style from OpenSCAD. > > I don't know how they handle the cube(a+b) case, though. My guess is that > they apparently-randomly pick one or the other variable to change, as long > as all of the rest of the constraints are still satisfied. > >
JB
Jordan Brown
Fri, Oct 23, 2026 11:33 PM

Constraint-based systems are clearly not BelfrySCAD, so I changed the
subject.

Moreover, while I agree that constraint-based systems are interesting...
OpenSCAD isn't one of them, so much more than a reference to "go look
over there for more information" is probably off topic.

But if people still want to talk about this, well, at least now the
subject is sort of appropriate.

Constraint-based systems are clearly not BelfrySCAD, so I changed the subject. Moreover, while I agree that constraint-based systems are interesting... OpenSCAD isn't one of them, so much more than a reference to "go look over there for more information" is probably off topic. But if people still want to talk about this, well, at least now the subject is sort of appropriate.
TA
Todd Allen
Sat, Oct 24, 2026 1:50 AM

OpenSCAD doesn't have a constraint system but there is nothing preventing
it but lack of developer capacity.  Imagine  existing language primitives
taking a ??? instead of a floating point value.  And ways to specify
constraints. The constraint solver keeps you informed of which ??? are
constrained and solvable or when constraints conflict.  If there is a GUI
things unconstrained can be dragged and then their distance or angle to
something else can be captured as a constraint.

BelfrySCAD could very likely leverage the Freecad sketcher into an
additional way to create 2d geometry in a couple weeks and expand it to 3d
quickly after that.  For OpenSCAD I think you would be right to say it is
impossible.

On Fri, Oct 23, 2026, 6:34 PM Jordan Brown via Discuss <
discuss@lists.openscad.org> wrote:

Constraint-based systems are clearly not BelfrySCAD, so I changed the
subject.

Moreover, while I agree that constraint-based systems are interesting...
OpenSCAD isn't one of them, so much more than a reference to "go look over
there for more information" is probably off topic.

But if people still want to talk about this, well, at least now the
subject is sort of appropriate.


OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD doesn't have a constraint system but there is nothing preventing it but lack of developer capacity. Imagine existing language primitives taking a ??? instead of a floating point value. And ways to specify constraints. The constraint solver keeps you informed of which ??? are constrained and solvable or when constraints conflict. If there is a GUI things unconstrained can be dragged and then their distance or angle to something else can be captured as a constraint. BelfrySCAD could very likely leverage the Freecad sketcher into an additional way to create 2d geometry in a couple weeks and expand it to 3d quickly after that. For OpenSCAD I think you would be right to say it is impossible. On Fri, Oct 23, 2026, 6:34 PM Jordan Brown via Discuss < discuss@lists.openscad.org> wrote: > Constraint-based systems are clearly not BelfrySCAD, so I changed the > subject. > > Moreover, while I agree that constraint-based systems are interesting... > OpenSCAD isn't one of them, so much more than a reference to "go look over > there for more information" is probably off topic. > > But if people still want to talk about this, well, at least now the > subject is sort of appropriate. > > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org
RD
Revar Desmera
Sat, Oct 24, 2026 3:49 AM

On Oct 23, 2026, at 6:50 PM, Todd Allen via Discuss discuss@lists.openscad.org wrote:

OpenSCAD doesn't have a constraint system but there is nothing preventing it but lack of developer capacity.  Imagine  existing language primitives taking a ??? instead of a floating point value.  And ways to specify constraints. The constraint solver keeps you informed of which ??? are constrained and solvable or when constraints conflict.  If there is a GUI things unconstrained can be dragged and then their distance or angle to something else can be captured as a constraint.

While I know how to make a constraints solver, it is far less clear how to integrate the necessary concepts into OpenSCAD syntax.
This is one area where the OpenSCAD Language's parent-child structure and implicit union-ing syntax makes things difficult.

You could do:
constraint_tangent() {
circle(20);
square([25,30]);
}

But that's insufficient constraints for that geometry.  What side of the square is it tangent to?  Which position coordinates are locked down?  And there's no clear way to specify multiple constraints for any one object.  If we had module literals, like PR #4478, we could theoretically add multiple constraints by referring to the shapes by variables, but even that gets muddied by the immutable "variables" Functional Programming.  Maybe something like:

a = module() square([25,30]);
b = module() circle(20);
a2 = add_constraint(a,"position",[0,0]);
b2 = add_constraint(b,"x>=",25);
b3 = add_constraint(b2,"y",0);
c = add_joint_constraint(a,b,"tangent");
c(); // instantiate!

You'll end up with a bunch of constraint types, and joint constraint types.

I'm not sure I want to really think about how to write the solver Jacobians for 3D tangents of irregular polyhedrons, though.

  • Revar
> On Oct 23, 2026, at 6:50 PM, Todd Allen via Discuss <discuss@lists.openscad.org> wrote: > > OpenSCAD doesn't have a constraint system but there is nothing preventing it but lack of developer capacity. Imagine existing language primitives taking a ??? instead of a floating point value. And ways to specify constraints. The constraint solver keeps you informed of which ??? are constrained and solvable or when constraints conflict. If there is a GUI things unconstrained can be dragged and then their distance or angle to something else can be captured as a constraint. While I know how to make a constraints solver, it is far less clear how to integrate the necessary concepts into OpenSCAD syntax. This is one area where the OpenSCAD Language's parent-child structure and implicit union-ing syntax makes things difficult. You could do: constraint_tangent() { circle(20); square([25,30]); } But that's insufficient constraints for that geometry. What side of the square is it tangent to? Which position coordinates are locked down? And there's no clear way to specify multiple constraints for any one object. If we had module literals, like PR #4478, we could theoretically add multiple constraints by referring to the shapes by variables, but even that gets muddied by the immutable "variables" Functional Programming. Maybe something like: a = module() square([25,30]); b = module() circle(20); a2 = add_constraint(a,"position",[0,0]); b2 = add_constraint(b,"x>=",25); b3 = add_constraint(b2,"y",0); c = add_joint_constraint(a,b,"tangent"); c(); // instantiate! You'll end up with a bunch of constraint types, and joint constraint types. I'm not sure I want to really think about how to write the solver Jacobians for 3D tangents of irregular polyhedrons, though. - Revar
RD
Revar Desmera
Sat, Oct 24, 2026 3:59 AM

And about the second after I hit send on this, it occurs to be that a pure-data constraints system is perfectly possible without module literals.

  • Revar

On Oct 23, 2026, at 8:49 PM, Revar Desmera revarbat@gmail.com wrote:

On Oct 23, 2026, at 6:50 PM, Todd Allen via Discuss discuss@lists.openscad.org wrote:

OpenSCAD doesn't have a constraint system but there is nothing preventing it but lack of developer capacity.  Imagine  existing language primitives taking a ??? instead of a floating point value.  And ways to specify constraints. The constraint solver keeps you informed of which ??? are constrained and solvable or when constraints conflict.  If there is a GUI things unconstrained can be dragged and then their distance or angle to something else can be captured as a constraint.

While I know how to make a constraints solver, it is far less clear how to integrate the necessary concepts into OpenSCAD syntax.
This is one area where the OpenSCAD Language's parent-child structure and implicit union-ing syntax makes things difficult.

You could do:
constraint_tangent() {
circle(20);
square([25,30]);
}

But that's insufficient constraints for that geometry.  What side of the square is it tangent to?  Which position coordinates are locked down?  And there's no clear way to specify multiple constraints for any one object.  If we had module literals, like PR #4478, we could theoretically add multiple constraints by referring to the shapes by variables, but even that gets muddied by the immutable "variables" Functional Programming.  Maybe something like:

a = module() square([25,30]);
b = module() circle(20);
a2 = add_constraint(a,"position",[0,0]);
b2 = add_constraint(b,"x>=",25);
b3 = add_constraint(b2,"y",0);
c = add_joint_constraint(a,b,"tangent");
c(); // instantiate!

You'll end up with a bunch of constraint types, and joint constraint types.

I'm not sure I want to really think about how to write the solver Jacobians for 3D tangents of irregular polyhedrons, though.

  • Revar
And about the second after I hit send on this, it occurs to be that a pure-data constraints system is perfectly possible without module literals. - Revar > On Oct 23, 2026, at 8:49 PM, Revar Desmera <revarbat@gmail.com> wrote: > > >> On Oct 23, 2026, at 6:50 PM, Todd Allen via Discuss <discuss@lists.openscad.org> wrote: >> >> OpenSCAD doesn't have a constraint system but there is nothing preventing it but lack of developer capacity. Imagine existing language primitives taking a ??? instead of a floating point value. And ways to specify constraints. The constraint solver keeps you informed of which ??? are constrained and solvable or when constraints conflict. If there is a GUI things unconstrained can be dragged and then their distance or angle to something else can be captured as a constraint. > > While I know how to make a constraints solver, it is far less clear how to integrate the necessary concepts into OpenSCAD syntax. > This is one area where the OpenSCAD Language's parent-child structure and implicit union-ing syntax makes things difficult. > > You could do: > constraint_tangent() { > circle(20); > square([25,30]); > } > > But that's insufficient constraints for that geometry. What side of the square is it tangent to? Which position coordinates are locked down? And there's no clear way to specify multiple constraints for any one object. If we had module literals, like PR #4478, we could theoretically add multiple constraints by referring to the shapes by variables, but even that gets muddied by the immutable "variables" Functional Programming. Maybe something like: > > a = module() square([25,30]); > b = module() circle(20); > a2 = add_constraint(a,"position",[0,0]); > b2 = add_constraint(b,"x>=",25); > b3 = add_constraint(b2,"y",0); > c = add_joint_constraint(a,b,"tangent"); > c(); // instantiate! > > You'll end up with a bunch of constraint types, and joint constraint types. > > I'm not sure I want to really think about how to write the solver Jacobians for 3D tangents of irregular polyhedrons, though. > > - Revar > >
JB
Jordan Brown
Sat, Oct 24, 2026 4:26 AM

Could one design a constraint-based language and fit it into the
OpenSCAD UI and geometry engines?  Definitely.  Am I interested in
working on such a project?  No.

Could one augment the OpenSCAD language with constraint mechanisms, so
that they would somehow play nice together?  Maybe.  I'm dubious. 
Probably depends on what kind of integration you would expect.  Am I
interested in working on such a project?  No.

Could one design a constraint-based language and fit it into the OpenSCAD UI and geometry engines?  Definitely.  Am I interested in working on such a project?  No. Could one augment the OpenSCAD language with constraint mechanisms, so that they would somehow play nice together?  Maybe.  I'm dubious.  Probably depends on what kind of integration you would expect.  Am I interested in working on such a project?  No.
RD
Revar Desmera
Sat, Oct 24, 2026 9:32 AM

Okay, it looks like it is possible to make a pure user-space OpenSCAD constraints system.  Not sure how fast it'd be.
It'd look something like this:

res = constraint_solve(
object(  // Initial guesses
cp=[4,3],  // center point of circle
r=2,      // radius of circle
a=[0,0],  // Point A of triangle
b=[1,0],  // Point B of triangle
c=[0,1]    // Point C of triangle
),
function(o) [  // Constraints
// A.y == B.y
constraint_horizontal(o.a, o.b),
// norm(A-B) == 20
constraint_length(o.a, o.b, 20),
// norm(C-midpt(B,C)) == 15
constraint_length(o.c, (o.a+o.b)/2, 15),
// norm(A-C) == norm(B-C)
constraint_equal_length(o.a, o.c, o.b, o.c),
// Circle is tangent to AB
constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.b),
// Circle is tangent to AC
constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.c),
// Circle is tangent to BC
constraint_circle_tangent_to_line(o.cp, o.r, o.b, o.c),
// Circle is centered at [0,1]
constraint_coincident(o.cp, [0,1])
]
);

draw_line(res.a, res.b);
draw_line(res.b, res.c);
draw_line(res.c, res.a);
translate(rec.cp) circle(r=rec.r);

Kind of ugly, and unnecessarily verbose.  I have got all the functions undefined above written out, but it's a few hundred lines long.

Honestly, I don't really know how often I'd want to go to all the trouble of formalizing constraints like this.

  • Revar
Okay, it looks like it *is* possible to make a pure user-space OpenSCAD constraints system. Not sure how fast it'd be. It'd look something like this: res = constraint_solve( object( // Initial guesses cp=[4,3], // center point of circle r=2, // radius of circle a=[0,0], // Point A of triangle b=[1,0], // Point B of triangle c=[0,1] // Point C of triangle ), function(o) [ // Constraints // A.y == B.y constraint_horizontal(o.a, o.b), // norm(A-B) == 20 constraint_length(o.a, o.b, 20), // norm(C-midpt(B,C)) == 15 constraint_length(o.c, (o.a+o.b)/2, 15), // norm(A-C) == norm(B-C) constraint_equal_length(o.a, o.c, o.b, o.c), // Circle is tangent to AB constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.b), // Circle is tangent to AC constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.c), // Circle is tangent to BC constraint_circle_tangent_to_line(o.cp, o.r, o.b, o.c), // Circle is centered at [0,1] constraint_coincident(o.cp, [0,1]) ] ); draw_line(res.a, res.b); draw_line(res.b, res.c); draw_line(res.c, res.a); translate(rec.cp) circle(r=rec.r); Kind of ugly, and unnecessarily verbose. I *have* got all the functions undefined above written out, but it's a few hundred lines long. Honestly, I don't really know how often I'd want to go to all the trouble of formalizing constraints like this. - Revar
RD
Revar Desmera
Sat, Oct 24, 2026 10:24 AM

That actually ran a lot faster than I thought it would. Converged in 7 iterations, in 0.012 seconds.  Maybe I should consider this for BOSL2.

  • Revar

On Oct 24, 2026, at 2:32 AM, Revar Desmera revarbat@gmail.com wrote:

Okay, it looks like it is possible to make a pure user-space OpenSCAD constraints system.  Not sure how fast it'd be.
It'd look something like this:

res = constraint_solve(
object(  // Initial guesses
cp=[4,3],  // center point of circle
r=2,      // radius of circle
a=[0,0],  // Point A of triangle
b=[1,0],  // Point B of triangle
c=[0,1]    // Point C of triangle
),
function(o) [  // Constraints
// A.y == B.y
constraint_horizontal(o.a, o.b),
// norm(A-B) == 20
constraint_length(o.a, o.b, 20),
// norm(C-midpt(B,C)) == 15
constraint_length(o.c, (o.a+o.b)/2, 15),
// norm(A-C) == norm(B-C)
constraint_equal_length(o.a, o.c, o.b, o.c),
// Circle is tangent to AB
constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.b),
// Circle is tangent to AC
constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.c),
// Circle is tangent to BC
constraint_circle_tangent_to_line(o.cp, o.r, o.b, o.c),
// Circle is centered at [0,1]
constraint_coincident(o.cp, [0,1])
]
);

draw_line(res.a, res.b);
draw_line(res.b, res.c);
draw_line(res.c, res.a);
translate(rec.cp) circle(r=rec.r);

Kind of ugly, and unnecessarily verbose.  I have got all the functions undefined above written out, but it's a few hundred lines long.

Honestly, I don't really know how often I'd want to go to all the trouble of formalizing constraints like this.

  • Revar
That actually ran a lot faster than I thought it would. Converged in 7 iterations, in 0.012 seconds. Maybe I should consider this for BOSL2. - Revar > On Oct 24, 2026, at 2:32 AM, Revar Desmera <revarbat@gmail.com> wrote: > > Okay, it looks like it *is* possible to make a pure user-space OpenSCAD constraints system. Not sure how fast it'd be. > It'd look something like this: > > res = constraint_solve( > object( // Initial guesses > cp=[4,3], // center point of circle > r=2, // radius of circle > a=[0,0], // Point A of triangle > b=[1,0], // Point B of triangle > c=[0,1] // Point C of triangle > ), > function(o) [ // Constraints > // A.y == B.y > constraint_horizontal(o.a, o.b), > // norm(A-B) == 20 > constraint_length(o.a, o.b, 20), > // norm(C-midpt(B,C)) == 15 > constraint_length(o.c, (o.a+o.b)/2, 15), > // norm(A-C) == norm(B-C) > constraint_equal_length(o.a, o.c, o.b, o.c), > // Circle is tangent to AB > constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.b), > // Circle is tangent to AC > constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.c), > // Circle is tangent to BC > constraint_circle_tangent_to_line(o.cp, o.r, o.b, o.c), > // Circle is centered at [0,1] > constraint_coincident(o.cp, [0,1]) > ] > ); > > draw_line(res.a, res.b); > draw_line(res.b, res.c); > draw_line(res.c, res.a); > translate(rec.cp) circle(r=rec.r); > > > Kind of ugly, and unnecessarily verbose. I *have* got all the functions undefined above written out, but it's a few hundred lines long. > > Honestly, I don't really know how often I'd want to go to all the trouble of formalizing constraints like this. > > - Revar > > >
JB
Jon Bondy
Sat, Oct 24, 2026 12:22 PM

"Honestly, I don't really know how often I'd want to go to all the
trouble of formalizing constraints like this."

That.

On 10/24/2026 6:24 AM, Revar Desmera via Discuss wrote:

That actually ran a lot faster than I thought it would. Converged in 7
iterations, in 0.012 seconds.  Maybe I should consider this for BOSL2.

  • Revar

On Oct 24, 2026, at 2:32 AM, Revar Desmera revarbat@gmail.com wrote:

Okay, it looks like it is possible to make a pure user-space
OpenSCAD constraints system.  Not sure how fast it'd be.
It'd look something like this:

 res = constraint_solve(
           object(  // Initial guesses
 cp=[4,3],  // center point of circle
               r=2,   // radius of circle
               a=[0,0],   // Point A of triangle
               b=[1,0],   // Point B of triangle
               c=[0,1]    // Point C of triangle
           ),
           function(o) [  // Constraints
               // A.y == B.y
 constraint_horizontal(o.a, o.b),
               // norm(A-B) == 20
 constraint_length(o.a, o.b, 20),
               // norm(C-midpt(B,C)) == 15
 constraint_length(o.c, (o.a+o.b)/2, 15),
               // norm(A-C) == norm(B-C)
 constraint_equal_length(o.a, o.c, o.b, o.c),
               // Circle is tangent to AB
 constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.b),
               // Circle is tangent to AC
 constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.c),
               // Circle is tangent to BC
 constraint_circle_tangent_to_line(o.cp, o.r, o.b, o.c),
               // Circle is centered at [0,1]
 constraint_coincident(o.cp, [0,1])
           ]
       );

 draw_line(res.a, res.b);
 draw_line(res.b, res.c);
 draw_line(res.c, res.a);
 translate(rec.cp) circle(r=rec.r);

Kind of ugly, and unnecessarily verbose.  I have got all the
functions undefined above written out, but it's a few hundred lines long.

Honestly, I don't really know how often I'd want to go to all the
trouble of formalizing constraints like this.

  • Revar

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

--
This email has been checked for viruses by AVG antivirus software.
www.avg.com

"Honestly, I don't really know how often I'd want to go to all the trouble of formalizing constraints like this.*"* That. * * On 10/24/2026 6:24 AM, Revar Desmera via Discuss wrote: > That actually ran a lot faster than I thought it would. Converged in 7 > iterations, in 0.012 seconds.  Maybe I should consider this for BOSL2. > > - Revar > > >> On Oct 24, 2026, at 2:32 AM, Revar Desmera <revarbat@gmail.com> wrote: >> >> Okay, it looks like it *is* possible to make a pure user-space >> OpenSCAD constraints system.  Not sure how fast it'd be. >> It'd look something like this: >> >> res = constraint_solve( >>           object(  // Initial guesses >> cp=[4,3],  // center point of circle >>               r=2,   // radius of circle >>               a=[0,0],   // Point A of triangle >>               b=[1,0],   // Point B of triangle >>               c=[0,1]    // Point C of triangle >>           ), >>           function(o) [  // Constraints >>               // A.y == B.y >> constraint_horizontal(o.a, o.b), >>               // norm(A-B) == 20 >> constraint_length(o.a, o.b, 20), >>               // norm(C-midpt(B,C)) == 15 >> constraint_length(o.c, (o.a+o.b)/2, 15), >>               // norm(A-C) == norm(B-C) >> constraint_equal_length(o.a, o.c, o.b, o.c), >>               // Circle is tangent to AB >> constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.b), >>               // Circle is tangent to AC >> constraint_circle_tangent_to_line(o.cp, o.r, o.a, o.c), >>               // Circle is tangent to BC >> constraint_circle_tangent_to_line(o.cp, o.r, o.b, o.c), >>               // Circle is centered at [0,1] >> constraint_coincident(o.cp, [0,1]) >>           ] >>       ); >> >> draw_line(res.a, res.b); >> draw_line(res.b, res.c); >> draw_line(res.c, res.a); >> translate(rec.cp) circle(r=rec.r); >> >> >> >> Kind of ugly, and unnecessarily verbose.  I *have* got all the >> functions undefined above written out, but it's a few hundred lines long. >> >> Honestly, I don't really know how often I'd want to go to all the >> trouble of formalizing constraints like this. >> >> - Revar >> >> >> > > > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email todiscuss-leave@lists.openscad.org -- This email has been checked for viruses by AVG antivirus software. www.avg.com