Re: Fwd: Data uri origin inheritance

On Thu, Jan 16, 2014 at 5:27 PM, Boris Zbarsky <[email protected]> wrote:
> On 1/16/14 10:21 AM, João Eiras wrote:
>>
>> The opener.
>
>
> Just to check, what makes you think there is an opener, in general? Consider
> the user opening a new tab, contains window W.  Then W opens a new window V.
> Then V navigates W to a data: URI.  W has no opener.
>

There is an opener when window.open() is called. There is a owner
document when using frames. But you know this already. I don't
understand why you are asking.

>
>> Apparently not. Testcase
>>
>> http://dl.dropboxusercontent.com/u/9681493/data_uri_20140116.html
>>
>> Is this valid ?
>
>
> I have no idea what this testcase is trying to do, at first glance, and what
> your expected and observed behavior are.  Care to try to explain?
>

The testcase does exactly what we've discussed in this thread, and it
expects a PASS.

> But note that origin and location.host don't match for data: URIs, for
> obvious reasons: location.host is "" but the origin is whatever it is. Not
> sure whether your testcase is expecting them to match, but I think it is...
>

It's code and a bit more than a page long, self contained, easy to
read. No, I didn't use location.host.
Now, I'd like someone to check if that's actually valid, because no
browser passes the testcase, that includes Firefox.