On 01/08/14 22:05, Josh Matthews wrote:
Can you elaborate on why browsing contexts are suspicious? I haven't
been able to come up with any reasons why out-of-process or out-of-task
would have a big impact here.
I don’t know if there really is an issue, I was asked that we look into
that:
http://krijnhoetmer.nl/irc-logs/whatwg/20140801#l-194
Each iframe has its browsing context. Since other browsers run iframes
in the same thread as their parent, they may not realize if some detail
of the spec *requires* everything to be in the same thread.
I only started looking at Fullscreen today, but here is what I gather:
Since full-screened elements can be nested (for example, a video
embedded in slides), there is a "fullscreen element stack".
When full-screening an element that is in an iframe, we want it to
occupy the entire (full) screen, not just that iframe. Therefore, only
top-level browsing context have a stack, which is shared with iframes.
There seem to be APIs (available to iframes) that want to determine
synchronously whether the stack (at the top-level) is empty.
The question is: can we do all this with parallel iframes? It might
involved blocking script execution while waiting for a response from the
script task in the top-level browsing context, which might be busy doing
something else.
Maybe it’s not actually a problem: we already block script in
getClientRects() and friends while waiting for results from the layout task.
Please review carefully, and decide if we should push for the spec to
change in some way.
--
Simon Sapin
_______________________________________________
dev-servo mailing list
dev-servo@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-servo
