resize node auto feature

The current resize node does not implement OpenSCAD's auto feature. I have extended it to do so. This allows dimensions to be unchanged, set to specific sizes or scaled proportionally to some other dimension.
For instance, to set an object's x extent to 15, scale in y proportionally to that, leave z unchanged, you resize([15,0,0],auto=[true,true,false])

It is well described in the OpenSCAD manual. I have tried to put a brief description in the hover text.
I have also added a second line of description to the scale node to make clear that it scales by specified scale factors, whereas resize sets absolute size.
I have no idea what the world/object/local flag does.

ok. will fix that and check about local and object space.

theoricaly, a cube rotated of 45 deg in Z axis and in world space should be scalable in local x axis.
Look at picture : world scale after rotation give different result than scale in object space after rotation.

This is true for each transfom, and very powerfull (in multicopter example, i can 'post' tilt motors with local rotation, without worrying of previous tranforms (rotations of arms around center).
with openscad, all tranforms are always in 'world' space.... and it leads to headache when you have a cascade of previous heterogeneous transforms (rotation and translations).

To be able to have this functionnality, graphscad always generate code to be able to get current transform matrix (4x4) of an object node in the graph.

It allows creation of nodes like 'lookat', that deal with previous transformation to make a correct aiming, even if you object is rotated, and not in center of world.

when you make some joining operations (typicaly a boolean), current matrix of output node will be the matrix of first input object connected to the node.

Interesting explanation. So, these spaces are to allow you to operate on an object as though it was still in its original orientation. I thought they must implement a feature of OpenSCAD that I had not learned, but actually it is a new feature you are adding on top of OpenSCAD. I guess this explains all the matrix functions that GraphSCAD creates.
I haven't looked at lookat. I will do so.

yes. this is a feature asked in openscad for long now. but still not present. It's not possible just cause openscad can't do it cause it's not really a procedural language : all is globally evaluated, and it s not possible to get intemediate results to use them. That why variable assignation is so hard to understand in openscad.