Preliminary Snapping when Dragging and Dropping

I worked on this several times over the last few days, but I didn't get to a point where it was working well enough to show until now...

I've been implementing the behaviour for snapping while dragging.

So, the basic situation is there's two sets of snap-points:

1. moving ones to snap with
2. stationary ones to snap to

and my first thought was get all of both sets, transformed into world coordinates, and snap when the i) the position and ii) the orientation were both within a threshold.  However, before implementing it I had RETHINK #1 and decided that just matching the position was better, and always snapping to the correct orientation, however far out it currently was.  This felt like it would be better than having to sequentially match up the rotation and the position.

This took longer than I expected, for one awkward reason which I had not considered when starting out.  The sheets are Godot Panels, which are a type of control, rather than a Node2D:

IDK whether this would have been easier for a Node2D-derived class, but I suspect it would, because moving and spinning is much more of an expected thing for those as they are intended much more as world components rather than UI components.  Anyway, the problem I had, was that for snapping to work, I need to track two positions and rotations for the moving sheet:

• the drag position, and.
• the display position.

Why two?  Because when we hit a snap location, we need to adjust the display position to show we have locked on to it, but we need the drag position to remain undamaged by that, in case this is not the snap-point that the user wanted.  This is especially the case when allowing arbitrary amounts of rotation snap.  Having the sheet leap around just because you passed something you aren't interested in is hardly the best user experience...

So I set off to read the transform off the sheet, drag that around and show the moving snap-points based on it.  While we aren't on a snap-point we just write that straight back and keep the display and drag positions the same.  When a moving snap-point is within the threshold of a stationary one, however, we compute the transform required to superimpose them exactly, and apply that to the moving sheet.  In position and rotation.

After a bit of to and fro, the recipe for that works at:

1. calculate the difference in orientation from the moving SP to the stationary one
2. add that to the moving sheet (the subtlety here is that the SP can be rotated inside the sheet, but the change in rotation needed for the SP in isolation is the same as that needed for the sheet to give the SP the same final orientation
3. using this new sheet transform, recalculate the SP position
4. calculate the delta from that to the stationary SP position
5. and add that to the sheet's position too

Simple, but it took me a while to get my head around what was actually required.

Got to go swim now, TBC...