In my opinion your approach makes completely no sense. You should have items for nodes and separate items for curves (or lines) between nodes and manipulate curves using those node items. But you can't implement it in 10 lines of code no matter how much you try it. Trying to detect control points for curves by observing coordinates is a totally bogus approach in an object oriented system such as Graphics View.
Don't take it personal but it's more and more often situation that I'm surprised people start programming without thinking first how they are going to implement the whole solution (in terms of designing the architecture, not writing the actual code) and they end up trying to implement some wacky solutions instead of using standard approaches used by everyone else in the industry. For instance -- how did you plan to move the control point in the path to some other position?
Bookmarks