For scenex to be usable in an application like pyapp-kit/ndv, we require the ability to modify model parameters based on user interactions (minimually the mouse buttons & scroll wheel). Ideally, this would be as simple as the scenex user saying "when the mouse does X (around my node Y), run my function Z", e.g.:

• When the mouse hovers over an Image node, we display (somewhere) the value under the mouse.
• When the mouse is pressed and then dragged, we translate the Camera node, unless the mouse is above another Node object - in that case we translate the object.

To provide an extensible mechanism enabling this behavior, we require an event abstraction that users can listen to. The questions, then, are?

• How are these events generated?
  • Do they arise from within the Canvas adaptors (i.e. harnessing native canvas backend events)?
  • Are they provided to the Canvas model (i.e. intercepted from the gui backend before they reach the canvas)?
• How might the user listen for an event?
  • Connect to signals on individual Nodes?
  • Interact with some independent structure?
• How do we implement default behavior?
  • How do we block default behavior when user behavior should take precedence?