Bug 44794 – Make sure rendering WebGL context inside a Canvas 2D context works

• Status: RESOLVED
• Resolution: WORKSFORME
• Component: WebGL

[Chris Marrin]

I believe all the logic is in place, but we need layout tests to prove this works

[Kenneth Russell]

There is already at least one layout test for this: fast/canvas/webgl/canvas-test.html .

[Chris Marrin]

(In reply to comment #1)
> There is already at least one layout test for this: fast/canvas/webgl/canvas-test.html .

Yes, you're right. I read this test wrong when I was looking for a test case. But we still need some additional tests. In particular, when Hyatt did his ImageBuffer change this test stopped working correctly (the blue box was showing up white), but the test still passed because it is not a pixel test. 

We need at least 4 pixel tests in addition to this one. They should match the css-webkit-canvas* tests I recently checked in:

1) Static WebGL rendering, put that on a 2D Canvas

2) Same as 1 but 100ms later render WebGL again (with a different background color) and put that on the 2D canvas. This tests that we can replace the image source.

3) Static 2D canvas rendering, put that on WebGL canvas

4) Same as 3, but 100ms later render 2D canvas again (with a different background color) and put that on the WebGL canvas.

We have similar tests to 3 and 4 already, but these are also not pixel tests. We want to minimize the number of pixel tests since they require platform specific results, but we do need some.

[Kimmo Kinnunen]

I think tests discussed here are already implemented.