Re: Sort out active state spec inconsistencies

I forgot that I already created an issue which links to this discussion.

/Adam

On 2015-12-01 15:14, Harald Alvestrand wrote:
> Den 01. des. 2015 11:03, skrev Adam Bergkvist:
>> Anyone else that has input on this matter?
>>
>> Otherwise, I'll go ahead and create an github issue.
>
> I'm fine with creating an issue and suggesting #3 as the resolution.
>
>>
>> /Adam
>>
>> On 2015-11-24 09:34, Adam Bergkvist wrote:
>>> I wouldn't mind going down the path of #3 since it would indeed simplify
>>> things to get rid of the double events. I wonder how much code we would
>>> break (besides browser layout tests).
>>>
>>> /Adam
>>>
>>> On 2015-11-23 23:18, Peter Thatcher wrote:
>>>> I like #3.  It simplifies.  The JS can get to the information it needs
>>>> and doesn't need it to be duplicated in a complex way.
>>>>
>>>> On Tue, Nov 17, 2015 at 2:18 AM, Adam Bergkvist
>>>> <adam.bergkvist@ericsson.com <mailto:adam.bergkvist@ericsson.com>> wrote:
>>>>
>>>>       Hi
>>>>
>>>>       We have some inconsistencies in the spec regarding the active state.
>>>>
>>>>       We define active in terms of the stream's track set. In short, if a
>>>>       non-ended track exists, the stream is active, otherwise inactive.
>>>>
>>>>       The active attribute is specified to return true if the stream is active
>>>>       and false otherwise.
>>>>
>>>>       We also specify that when a stream goes from active, to inactive, the
>>>>       active attribute should be updated to true and the event fired. This
>>>>       contradicts the other text that defines the attribute in terms of the
>>>>       stream's track set.
>>>>
>>>>       Possible solutions:
>>>>
>>>>       1. Just remove the attribute updating part of the event dispatching
>>>>       steps.
>>>>
>>>>       Since the events are scheduled and fired as a result of sync API calls,
>>>>       we might en up in a situation that an "active" event is fired, but
>>>>       that's not true (anymore).
>>>>
>>>>       2. Remove the attribute updating part of the event dispatching and use a
>>>>       activitychanged event instead of specific active/inactive events.
>>>>
>>>>       This way when a scheduled event fires, the script needs to check the
>>>>       attribute manually and react. We never get the "event type" - "attribute
>>>>       value" mismatch.
>>>>
>>>>       3. Remove the active/inactive events all together.
>>>>
>>>>       These events are rather unnecessary. They either follow direct
>>>>       manipulation from the script or some other event when the UA has
>>>>       initiated a the result of a change to a stream's track set. For example
>>>>       the add/removetrack events or MediaStreamTrack ended.
>>>>
>>>>       /Adam
>>>>
>>>>
>>>
>>>
>>>
>>
>>
>
>


Received on Tuesday, 1 December 2015 15:46:47 UTC