ok- I understasnd what you say - I personally have not yet done it all the way. I met somebody internetwise, who told me that he needed my pixel drawing that I had created for printing out and creating templates in vektor (*.DXF) in order for his cnc ...
So I got started redrawing in vektor and other then you, I have all layers in one file (and in *.dxf) now.
The thing that he does on the machine is called "extrusion" in the 3D modelling language - he extrudes the 2D file to its 3D depth/hight whatever, it is a very simple parametric numeric input (z-axis)
He still needs the information on how deep each extrusion wants to be - in 2D that is a matter of extra written information, while the 3D model should read it out.
Looking at it from here I actually think, that the single file way you discribe is better when you have many layers, because you can name them better, the layer/object name menu is not the best in the *.dxf I think - its not like a photoshop file the name might not be maintained between different programs. That is a problem when you have a file and a routing that is called e.g: " stealth board under pickguard route" - I mean unusual extra things in uncommon places
I'd be interessting to read what the opener of this thread has to say, how he preferes
format and files ?
The roundover, so I imagine, is only done from the shape and the use of the router bit along the line. That is what I ment with communication or 3 D model - only the 3D model can tell - "ah that edge is a roundover and the other is for a binding".
I don't even know if the complete 3D model would get confusing since you don''t see so clear in which step you have to change the bit - these things I would like to clear - no machine will tell us " I know what you mean" unless you first tell it