Leaderboard

Nice work. I built the same kind of wheel for Revit a while back, so it was good to see someone doing it on the AutoCAD side. I could not see it running from the thread either, so a short screen recording would probably get you more replies here than anything else. Everything below is from my side, on Revit, so translate it however it fits. These are the ones that cost me the most time. Aim by the pointer angle, not by which shape is under the pointer. Point in a direction and that wedge lights up, whether you are near the middle or way out past the rim. That one change is what made mine feel like a game wheel instead of a menu. Number keys 1 to 9 pick a slot directly. Once you know your own layout you stop looking at the wheel at all, and that is where the real speed comes from. Grey out a slot that cannot run right now, and work all of them out in one pass before the wheel opens. Mine has to check the current selection to answer that. Doing it slot by slot meant a 50,000 element selection got walked once for every slot on the wheel. Once for the whole wheel fixed it. Check the command will really be accepted before you fire it. When mine was refused, nothing ran but the wheel still closed as if it had worked. That is the worst kind of failure because it looks like an answer. Smaller things I would not have guessed: Test the wedge maths outside CAD. I rebuilt mine as a standalone renderer and produced every slot count at every wheel size as images. Far quicker than restarting the host for each attempt, and it found a label collision at my highest slot count that I would never have caught by eye. Store the command name in the slot, not the button text. Rename a button later and the layout survives. Slot 1 at the top, the rest clockwise, always. The muscle memory is the whole point, so that order can never move. Drop any glow or fade. Mine had a soft shadow on the hovered wedge and a fade across the whole ring, and both were being recomputed every time the pointer crossed a boundary. A brighter fill and a thicker outline read exactly the same and cost nothing. Only redraw the wedge that changed, never the whole ring. Let the config file be missing or corrupt without complaining. Mine falls back to the defaults instead of throwing a message at me, and I have never once wanted the message. Do not let dead keys close it. Enter, space and a number higher than my slot count all used to dismiss the wheel doing nothing, which just felt broken. Open the settings from the wheel itself, and let people set it up by dragging onto a real wheel instead of filling in a list. Drag a tool onto a slot to fill it, drag one slot onto another to swap them. Mine draws the same ring at a smaller size for that, so what you arrange is what opens. Two things I am curious about, since you draw with real entities in model space rather than a window over the top. Does the wheel stay the same size on screen at any zoom, or does it scale with the drawing? Mine sits in a window above everything so I never had to deal with that, but it looks like the harder half of your approach. And the double click. In AutoCAD a double click on an object opens its editor, so what happens when the pointer is over an entity instead of empty space? I went with a keyboard shortcut and dodged the question, but yours is the nicer gesture if it works.

That's a point. Making the drawing as simple, as possible can definitely stop issues from happening when the client gets the DWG file. Using fonts not adding extra things and handling external references the right way can make a big difference. I have also noticed that missing supporting files are often the reason a drawing works differently on another computer.

If you use Etransmit it will add extra files to what is exported, in particular say custom fonts and xrefs.