Support zero dimension values for aspect-ratio

[annevk]

Also map to aspect-ratio consistently with how the attributes are parsed.

Tests: web-platform-tests/wpt#28229.

Follow-up: #6528.

Closes #4961 and closes #6527.

• At least two implementers are interested (and none opposed):
  • Google (presumably)
  • Mozilla
• Tests are written and can be reviewed and commented upon at:
  • [Gecko Bug 1700640] Map width and height to aspect-ratio in <canvas>, <input type=image>, and <video>. web-platform-tests/wpt#28229
  • HTML: update aspect-ratio rendering tests web-platform-tests/wpt#28932
• Implementation bugs are filed:
  • Chrome: https://bugs.chromium.org/p/chromium/issues/detail?id=1207330
  • Firefox: N/A
  • Safari: https://bugs.webkit.org/show_bug.cgi?id=225648

(See WHATWG Working Mode: Changes for more details.)

---

/canvas.html ( diff )
/embedded-content.html ( diff )
/grouping-content.html ( diff )
/infrastructure.html ( diff )
/interactive-elements.html ( diff )
/media.html ( diff )
/obsolete.html ( diff )
/rendering.html ( diff )

[domenic]

Ping @cbiesinger @emilio for their subject-expert review; I can do editorial review but I apparently didn't spot previous flaws in this area so your help would be much appreciated to make sure we get it right this time :)

[annevk]

@cbiesinger @emilio ping.

[emilio]

Looks good to me.

[annevk]

I created web-platform-tests/wpt#28932 to update the tests and filed https://bugs.chromium.org/p/chromium/issues/detail?id=1207330 against Chrome.

Safari has a lot more failures, despite https://bugs.webkit.org/show_bug.cgi?id=201641 being marked fixed. Is that expected @smfr? (Edit: filed https://bugs.webkit.org/show_bug.cgi?id=225648 meanwhile.)

[domenic]

Changes LGTM, would be good to clean up the source paragraph while you're there.

[annevk]

That sentence is rather weird. I wish we had made that flow from the img element itself (as the picture element works in general), rather than in this roundabout way. As in, it seems this should have been a change to the source set to also list these sizes.

[domenic]

My impression as to why this sentence is a bit weird is that it's a statement of fact about the normative requirements in the rendering section. Since it's in the source section, it's source-element centric and not super-precise.

[zcorpan]

@domenic yep. It could technically be removed altogether without changing what the spec requires, and so from that perspective there's flexibility in how to word it. It could also be changed to a note to further signal its non-normative nature.