Struct Canvas 

pub struct Canvas {
    rect: Rect,
    response: Response,
    diagram: ShapeIdx,
    pan: Camera,
    moves: Vec<Move>,
}

One canvas frame as the pointer left it: the rect it was laid out in, the moves it read for the session to apply, and the raw input the session resolves. A value rather than a borrow of the view, so the frame can call the editor between reading its input and painting.

Fields

rect: Rect

response: Response

diagram: ShapeIdx

Where the diagram is painted, reserved as the canvas is laid out.

pan: Camera

moves: Vec<Move>

Implementations

impl Canvas

pub fn viewport(&self) -> Rect

The screen rect this frame paints into.

pub fn moves(&mut self) -> Vec<Move>

What the pointer did to the camera this frame — a drag pan, a wheel notch, a pinch — in the order it happened, for the session to apply before the pointer is resolved.

pub fn input(&self, ui: &Ui, glass: &[Rect]) -> Input

What the pointer and the keys did this frame, as the core resolves them: the pointer’s motions and button edges while it is over the canvas or holding a press that began there, its leaving otherwise, and a pan in progress, which owns the pointer.

Presses that began on a toolbar button never arrive: holding one — whose screen point maps to some world position — must not arm a canvas affordance. egui keeps that promise for chrome in a layer of its own; for chrome the host draws in the canvas’s layer it counts the canvas as hovered on the press and not on the release, so the host names the rects of such chrome as glass, and a press on one is not taken either.