feat(CanvasContext): Added optional devicePixelRatio argument to resize

[saitonakamura]

It's mainly useful in context of rendering in web worker, because web worker doesn't have access to devicePixelRatio, so resize code will be split between

1. main thread asking for a render, providing devicePixelRatio
2. Worker rendering and resizing CanvasContext w/ provided ratio
3. Worker giving back main thread the resize target
4. Main thread applying css width/height provided by the worker

[rvilarl]

Do you use the SVGContext? Would something similar be required there?

[ronyeh]

Thanks!

I think we have been slowly going away from checks against null.
Instead we try to use the ?? operator where possible.

Does this code work for you?

  resize(width: number, height: number, devicePixelRatio?: number): this {
    const canvas = this.context2D.canvas;
    const dpr: number = devicePixelRatio ?? globalObject().devicePixelRatio ?? 1;

    // Scale the canvas size by the device pixel ratio clamping to the maximum supported size.
    [width, height] = CanvasContext.sanitizeCanvasDims(width * dpr, height * dpr);

    // Divide back down by the pixel ratio and convert to integers.
    width = (width / dpr) | 0;
    height = (height / dpr) | 0;

    canvas.width = width * dpr;
    canvas.height = height * dpr;

    // The canvas could be an instance of either HTMLCanvasElement or an OffscreenCanvas.
    // Only HTMLCanvasElement has a style attribute.
    if (isHTMLCanvas(canvas)) {
      canvas.style.width = width + 'px';
      canvas.style.height = height + 'px';
    }

    return this.scale(dpr, dpr);
  }

If it works for you, can you resubmit the PR with this approach? I'll be happy to merge it.

Thanks for your contribution!

[saitonakamura]

Do you use the SVGContext? Would something similar be required there?

Thanks for the feedback! I personally don't, but I don't think SVGContext takes any advantage of the devicePixelRatio, it's usually only the canvas thing in my experience. Also don't see it in the svgcontext code

I think we have been slowly going away from checks against null. Instead we try to use the ?? operator where possible.

Thanks for the feedback too! Works for me, updated

[rvilarl]

LGTM thanks.

[ronyeh]

Thanks!