Isn’t it premature to worry about documenting stuff hadn’t even made it
into dev yet?

On Fri, Aug 8, 2025 at 10:42 vulcan_--- via Discuss <
discuss@lists.openscad.org> wrote:

On 8/8/2025 7:41 AM, vulcan_--- via Discuss wrote:

Updating the docs is premature, since the feature isn't in yet.

The essence is the same as it is in JavaScript.

If you have an object o with a member f that is a function, and you
invoke the function as o.f(...), the function is provided with a special
variable $this that is the object.  (Or that is a copy of the object;
you can't tell.)

This allows you to do things like

function rect(size) = object(
     size = size,
     area = function() $this.size.x * $this.size.y
);
r = rect([10,20]);
echo(r.area());

Jordan Brown wrote:

all i needed to read was that it is premature to document it .. mind you, i did just see the PR pass all tests so ..

all done then, and no action to take

On 8/8/2025 5:10 PM, vulcan_--- via Discuss wrote:

Doesn't mean that it's in its final form. I've had several PRs pass all
of the tests and then die, or change radically, or go on hold for a long
time.

I move it out of draft and hope it gets accepted soon!

Basically a function in an object will know where it came from. The object it came from will be $this. Additionally, it can access all fields in the objects without this prefix.

o = object( 
	a	=	42, 
	f1	=	function() a, 
	f2	=	function() $this.a );

assert( o.f1() == 42 );
assert( o.f2() == 42 );

The context of a method (a function from an object) is:

parent context
fields from objects + $this
parameters
local variables

The lookup is for bottom to top. So a parameter that has the same name as a field will have priority. A local variable overrides the parameter and the object field. You can always access the field with the $this variable.

Functions remain bound tot the object unless assigned to another object.

f1 = o.f1;
assert( f1() == 42 );
f = object( a=4711, f=f1 ).f;
assert( f() == 4711 );

Functions have access to themselves so you can make recursive functions like in a Y combinator:

Y = object( f=function(n) n > 1 ? n*f(n-1) : 1).f; 
assert(Y(5)==120);

This will allow for objects. For example,

solid  = Prism( size=[10,10,10], size2=5);
inside = box.inside(wall=4);

As an old OO guy I think it might be useful to establish some conventions around some meta data for these objects (like a class name) but that has no urgency. Maybe best to wait a bit with this until we have more experience.

I hope the PR will soon be accepted. I am not unhappy with the current solution.

Peter Kriens