- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Wed, 19 Oct 2011 14:24:06 -0400
- To: public-web-perf@w3.org
On 10/19/11 2:12 PM, Boris Zbarsky wrote: > On 10/19/11 2:05 PM, Karen Anderson (IE) wrote: >> What I find interesting is that IE and Chrome always have >> loadEventStart == loadEventEnd in both the onload event and in the >> callback. > > Which exact load event are you observing in this case? That's not clear > from the script. I suspect that in IE and Chrome it matters Here's a question, actually. Does the "load" event on the <iframe> element fire synchronously with the load event on the window inside in Trident and WebKit? What about Presto? In Gecko, it does. The current HTML5 draft says that the load event on the iframe should fire off an async task that is posted when load event handlers have finished running on the subframe's window. If that's what WebKit and Trident are implementing, that explains the behavior differences on this test. It should also be detectable via other means (e.g. code would be able to run off a timeout before the iframe's onload but after the contained window's onload. But it's not clear to me why running this async is desirable and whether that part of HTML5 is stable enough that the navigation timing tests should be depending on it at this point. -Boris
Received on Wednesday, 19 October 2011 18:24:48 UTC