There are two changes here:
1) We allow setting .body even if the root element is not an <html:html>. This is what the spec says to do, and what we used to do before the changes in bug
366200. No tests for this yet, pending https://github.com/whatwg/html/issues/3403 getting resolved.
2) We use GetBody(), not GetBodyElement(), to look for an existing thing to replace. This matters if there are <frameset>s involved.
Tag UXP Issue #1344
Tag UXP Issue #252
Enable skipping loop filter and allow non spec compliant speedup tricks.
to enable, create new boolean pref "media.ffmpeg.skip_loop_filter" and set it to true.
CheckForOutdatedParent()
This was only used to check for cases when document.open() changed the
global, and elements being inserted into the document needing a new
reflector as a result.
Since document.open() no longer changes the global, this code is no
longer needed.
This changes the work we do for document.open() in the following ways:
- We no longer create a new Window when doing document.open().
We use the same Window but remove all the event listeners on the
existing DOM tree and Window before removing the document's existing
children to provide a clean slate document to use for .write().
- We no longer create a session history entry (previously would be a
wyciwyg URI). We now replace the current one, effectively losing the
entry for the original document.
- We now support document.open() on windowless documents.
https://bugzilla.mozilla.org/show_bug.cgi?id=1244768
I happened to find an older version of the promise-based media playback patch in Bugzilla, the one that was originally submitted for review. It had the DocShell changes I already knew how to deal with, and had fewer of the audio wrapper and nsISupports changes that were confusing me in the later patch. I was able to do a better job getting this back into a UXP-appropriate configuration than I could have with the final version.
I'm honestly still a little unsure about some of the minor refactoring done in the patch itself, insisting on already_AddRefed promises and such, but I don't really know how to avoid those completely. Still, I think it's better than it was.
This updates our behavior for computed DOM styling to no longer return
null on elements that have no display, but return a 0-length (empty)
style instead and don't throw. For this we stop looking at having a
presentation for the style and just look at the document instead.
This resolves#1219
