Re: Fwd: Data uri origin inheritance

On 1/16/14 12:07 PM, Boris Zbarsky wrote:
> If you want me to spend time answering that question, I really do
> suggest you do the above, because I simply don't have any more time to
> spend on this discussion if we're doing to ignore parts of it and write
> obfuscated code.

But ok, let me see if I'm following this mess right:

1)  The page is loaded from dl.dropboxusercontent.com.  It sets 
document.domain to itself.

2)  "start-navigation-frame" is loaded from foobar.dl.dropboxusercontent.com

3)  The frame sets document.domain to "dl.dropboxusercontent.com".

4)  The frame posts a message to itself and extracts the "origin" from 
the message event, whatever that is in this case.

5)  The frame the sets the .src on a frame in the parent, to a string 
that will run a script that expects the origin to be that of the frame.

Correct?

If so, no, that's not a valid testcase.  We talked about this up-thread 
already.  Just to refresh your memory, please read 
http://lists.w3.org/Archives/Public/www-dom/2014JanMar/0025.html and 
note that setting iframe.src and setting iframe.contentWindow.location 
have different behavior: the former is treated as a navigation by the 
document containing the iframe (because in fact that's what does the 
navigation), while the latter is treated as a navigation by the script 
doing the location set.  They differ not just in terms of origin but 
also base URI resolution, for example.

-Boris