On Tue, 17 Jun 2008 17:47:24 +0200, Boris Zbarsky <[EMAIL PROTECTED]> wrote:
Anne van Kesteren wrote:
The constructor and providing a pointer to the document from the object
on which the constructor was invoked.
So correct me if I'm wrong, but wouldn't it work to say that if the
object the constructor is on is a Window things work as now and
otherwise whatever specification placed the constructor on some other
object is responsible for describing how one gets a document from that
object?
I suppose. Though it would create some path in the XMLHttpRequest
specification we can't really test and it's not clear to me why we should
do that.
Also, I forgot to mention that the definition of origin in HTML5 depends
on Window and we really want to use the HTML5 definition of origin.
Note that a lot of the text can still be shared (in terms of whether the
calling scope or callee object determines the "right" document to use);
all that depends on Window is how one actually gets the document, right?
Actually, no, origin depends on it too. I forgot about that because
XMLHttpRequest just references origin directly.
Having basic Window support is also good for the test suite so we can
actually test the cross-frame scenarios you raised a long time ago (for
which I added this dependency when I addressed the issue).
I don't see a problem with saying that the test suite requires a Window,
to be honest. If we really wanted we could somehow flag the tests in
the test suite that really have that as a hard requirement, of course,
but realistically the tests are HTML files, aren't they? So they would
always have a Window...
That's fair enough.
--
Anne van Kesteren
<http://annevankesteren.nl/>
<http://www.opera.com/>
