This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The outcome of bug 21855 was to allow implementations to avoid firing a message event if there is already an in-memory session, possibly already containing a license, for the provided initData. This was actually implemented as part of the state machine [1], and a clarification was later added to indicate that the sessions must still appear distinct [2]. I think such behavior causes inconsistent and unexpected behavior for applications. I think EME should not allow such *optional* in-memory sharing. Instead, we should try to really solve the underlying problem (redundant needkey events) and/or make the behavior explicit (i.e. adding another value to bug 25200). The changes to the spec would be: A. Remove the note added in [2]. B. Require a message event when creating a session unless otherwise specified. There might be use cases for not firing a message when creating a session, but I don't think they are currently explicitly supported. We can re-evaluate the related algorithms when enabling such use cases. [1] https://dvcs.w3.org/hg/html-media/rev/83629aec22e1 [2] https://dvcs.w3.org/hg/html-media/rev/cba144274140
The underlying problem is tracked as bug 25268.
Also, as discussed at the time, it seems difficult to define or implement this behavior in a way that still allows the "MediaKeySession instances [to] appear distinct" as is stated in [2].
(In reply to David Dorwin from comment #2) > Also, as discussed at the time, it seems difficult to define or implement > this behavior in a way that still allows the "MediaKeySession instances [to] > appear distinct" as is stated in [2]. If your proposal for bug 25199 is accepted, it looks like this problem will go away.
(In reply to Joe Steele from comment #3) > (In reply to David Dorwin from comment #2) > > Also, as discussed at the time, it seems difficult to define or implement > > this behavior in a way that still allows the "MediaKeySession instances [to] > > appear distinct" as is stated in [2]. > > If your proposal for bug 25199 is accepted, it looks like this problem will > go away. Not necessarily. I was referring to the underlying state, not just the reporting of results. For example, if one session is release()'d, how does that affect the other session in a way that makes sense while still allowing the sessions to "appear distinct"?
The resolution of bug 25268 (all implementations check for existing sessions) addresses the original use case and makes the associated text obsolete. The text is removed in https://dvcs.w3.org/hg/html-media/rev/a91faf82b9b5