Fix createImageData with sizes < 1 pixel

[dzbarsky]

[pcwalton]

Is there a test for this?

[dzbarsky]

Yep, the one I deleted the metadata for :)

[jdm]

This seems odd to me. There's no explicit spec text for this, and Gecko doesn't appear to have a similar maximization step: https://dxr.mozilla.org/mozilla-central/source/dom/canvas/CanvasRenderingContext2D.cpp#5365

[jdm]

I'm going off of https://html.spec.whatwg.org/multipage/scripting.html#pixel-manipulation:dom-context-2d-createimagedata-3 by the way.

[dzbarsky]

Gecko does the equivalent thing here: https://dxr.mozilla.org/mozilla-central/source/dom/canvas/CanvasRenderingContext2D.cpp#5341 (this is after calling abs).

The relevant spec text is a little lower:
"New ImageData objects must be initialised so that their width attribute is set to the number of pixels per row in the image data, their height attribute is set to the number of rows in the image data...At least one pixel's worth of image data must be returned."
So the minimum size is 1x1.

[dzbarsky]

Unfortunately the canvas spec doesn't really list explicit steps for its algorithms, you have to infer them from various requirements. That's why we fail so many edge-case tests.

[jdm]

Delightful!

[jdm]

@bors-servo: r+

[bors-servo]

📌 Commit 2fb5fda has been approved by jdm

[bors-servo]

⌛ Testing commit 2fb5fda with merge 6c4d9ad...

[bors-servo]

💔 Test failed - linux1

[jdm]

@bors-servo: retry

• infra issue

[bors-servo]

⌛ Testing commit 2fb5fda with merge debd7d8...

[bors-servo]

☀️ Test successful - android, gonk, linux1, linux2, linux3, mac1, mac2, mac3