[Webkit-unassigned] [Bug 36465] window.Document !== window.XMLDocument

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=36465


Alexey Proskuryakov <ap at webkit.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|New Bugs                    |WebCore JavaScript
                 CC|                            |mjs at apple.com




--- Comment #25 from Alexey Proskuryakov <ap at webkit.org>  2010-03-23 17:00:19 PST ---
I don't think we'll want to introduce XMLDocument interface unless it's defined
in HTML5. We made much more drastic departures from current Firefox behavior to
match the agreed-upon behavior in HTML5 (e.g. the aforementioned change to give
HTML elements XHMTL namespace), and Firefox is changing in the same direction.

This is not really an aesthetic argument. The differences are very much
detectable from JS code. For example, suppose that one calls document.open() on
an SVG document. Per HTML5, this changes it into an HTML one - but it's still
the same object, and thus implements the same interfaces.

In any case, this is a discussion for WHATWG mailing list, if you want to bring
it up there.

> Do you have an alternative proposal for resolving Bug 9063?

One way to implement document.load without regressing behavior on sites that
call load() from attribute event listeners is to only let it be found in
specific cases. E.g., only via dot notation, and only on the object's prototype
itself. This could be done at property lookup time, or even in the JS parser.
I'm sure someone familiar with JSC would have more and better ideas.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.