[media-source] VideoPlaybackQuality is underdefined, leading to interop concerns

wolenetz has just created a new issue for 
https://github.com/w3c/media-source:

== VideoPlaybackQuality is underdefined, leading to interop concerns 
==
While looking into #69, we discovered some concerns with the current 
specification of VideoPlaybackQuality:
* The spec does not say what to do if values cannot be provided, 
potentially leading to interop issues and/or authors ignoring them.
 * For example: Firefox and Chrome always report corruptedVideoFrames 
as 0, meaning a value of 0 is meaningless to authors. If this value 
can’t be calculated by most implementations, maybe it should be 
removed from the spec. Otherwise, there should be a specification of 
how to do feature detection. For example, state that implementations 
should not expose an attribute it cannot support or otherwise indicate
 it is invalid (a nullable/optional value?).
 * If the implementations are inaccurate, it may be misleading 
authors.

Note: Moving VPBQ out of MSE spec (per #69), would address this MSE 
bug.

As further indication that this feature is underdefined:
* Is there no concern or parallel for corrupted __audio__ decode? 
(Perhaps this should be a MediaQualityObject.)
* droppedVideoFrames covers both predecode drop and missing deadline 
drop:
 * is this distinction relevant to authors, and can all/most 
implementations cover both types of drops?
 * If not, this value would be unreliable and we need some sort of 
feature detection.
* Optionality in whether to report corrupt frames as dropped: This 
allows implementation flexibility, but should implementations be 
consistent in reporting?


Please view or discuss this issue at 
https://github.com/w3c/media-source/issues/76 using your GitHub 
account

Received on Monday, 16 May 2016 21:54:19 UTC