- From: Philip Jägenstedt <philipj@opera.com>
- Date: Thu, 09 Apr 2009 17:27:37 +0200
- To: Philip Jägenstedt <philipj@opera.com>, public-html@w3.org
On Thu, 09 Apr 2009 10:49:33 +0200, Philip Jägenstedt <philipj@opera.com> wrote: > Some feedback on the big batch of changes to media elements made in > March. > > 1. may never load > > Previously there were many conditions for when to implicitly invoke the > load() method, e.g. when a media element was inserted into a document. > This was all replaced by the resource selection algorithm, but as far as > I can see that algorithm is only invoked by an explicit load() and in > the Audio(src) constructor. > > In other words, there's nothing to cause e.g. <video autobuffer > autoplay> to either begin buffering or playing unless a script > explicitly calls load() or play() on it. > > For simplicity I would prefer that the resource selection algorithm be > started for any media element, regardless of whether it has any > src/source, is in a document or any other conditions. This makes all of > the below equivalent: > > <audio src="foo"> > > a=new Audio("foo"); > > a=new Audio(); > a.src="foo"; > > a=document.createElement("audio"); > a.src="foo"; > > This behavior isn't incredibly important in itself, but it removes the > need for special cases in the Audio constructor or elsewhere. Ignoring > whether or not the media element is in a document when invoking the > resource selection/fetch algorithm makes sense for preloading, etc so I > don't think we should reintroduce that. My apologies, a text searching bug in Opera caused me to miss the following: If a media element whose networkState has the value NETWORK_EMPTY is inserted into a document, the user agent must invoke the media element's resource selection algorithm. In other words, my above complaint is invalid and I don't suggest any changes in how the resource selection algorithm is invoked. The examples given above will behave differently, but that's not a big problem as far as I can see. Hope I didn't overlook something with the other two points... -- Philip Jägenstedt Opera Software
Received on Thursday, 9 April 2009 15:27:36 UTC