Allow the 'root' argument to be a Document.

[szager-chromium]

Issue #372

Closes #???

The following tasks have been completed:

• Modified Web platform tests (link to pull request)

Implementation commitment:

• WebKit (https://bugs.webkit.org/show_bug.cgi?id=)
• Chromium (https://bugs.chromium.org/p/chromium/issues/detail?id=)
• Gecko (https://bugzilla.mozilla.org/show_bug.cgi?id=)

---

Preview | Diff

[bzbarsky]

Where is "the document's viewport" defined? Is it defined for documents that are not in a browsing context or are not fully active?

[szager-chromium]

This is sort of intentionally vague. The existing spec says this:

"If the intersection root is the implicit root, it [the root intersection rectangle] is the viewport’s size."

Elsewhere it mentions "the viewport of the document" and "viewport of the nested browsing context". It also says this in a highlighted note:

"Root intersection rectangle is not affected by pinch zoom and will report the unadjusted viewport, consistent with the intent of pinch zooming (to act like a magnifying glass and NOT change layout.)"

However, all of that text is likely to change with the resolution of this issue:

#95

The intention here is to use "viewport" in a way that is consistent with existing usage in the spec, until the above issue is resolved.

[bzbarsky]

I understand all that, but the implicit root, being the active document of a toplevel browsing context, always has at least some concept of a viewport. If there's some random document involved (e.g. an XHR responseXML document, or a document from an iframe that is now removed from the DOM or whatever), there may be no concept of a viewport at all. #95 does not cover that, as far as I can tell.

[szager-chromium]

The only spec language I could find that defines browser viewport is this unhelpful bit:

https://drafts.csswg.org/css2/visuren.html#viewport

But I'm not sure it's necessary to determine what the browser viewport of a detached document is, since the "Update the rendering" steps in the HTML event loop -- which is where IntersectionObserver runs -- only applies to fully active (connected) documents:

https://html.spec.whatwg.org/multipage/webappapis.html#event-loop-processing-model

[bzbarsky]

Aha. It might be good in the loop to assert that all documents we encounter are fully active. Those might still not have a sane viewport (e.g. if their embedding iframe is display:none), but at least it cuts down on the cases that need to be considered.

[szager-chromium]

OK, I added some language there to clarify that the root document must be fully active at this step in processing.

PTAL

[bzbarsky]

This looks reasonable to me, thank you.

[szager-chromium]

@bzbarsky Would you mind clicking the "approved" button...

[szager-chromium]

@bzbarsky If you could take another look at this soon-ish, that would be helpful. We're hoping to ship this behavior in the next branch of chromium, which will happen next week.

Thanks