Re: [rtcweb] Relaxing SDP O/A (was RE: Agenda requests for Atlanta meeting)

["Cullen Jennings (fluffy)" <[e-mail]>]

On Oct 19, 2012, at 3:23 AM, "Hutton, Andrew" <[e-mail]> wrote:

> 
> RFC3624 requires that the session reverts to the state prior to the offer if the offer/answer exchange fails. 
> 
> If an applications attempts to modify a session but the updated offer is rejected by the far end there should be no interruption to the existing session. Like Roman suggested I am also thinking it would be nice if the API provided some way of cancelling an offer and reverting to the original state but if this is putting too much state back in the browser we would need a clearly documented mechanism.

We talked about this before and I think just about everyone agrees that we need rollback in JSEP.  And it's not just needed for gatewaying with SIP, any signaling protocol that wanted to allow simultaneous changes is likely need it.  It's not yet fully defined in the current draft because we need to get stable on what we do the initial "O/A" pair before we step into the modifications of it. In SIP terms this would be let make sure we understand initial INVITE before taking on UPDATE. 

I agree this is 1) needed and 2) needs to be well defined in JSEP 3) it's not well defined in current draft. The idea/hope is that the current API will support rollback with no change to the API, we just have to write down the semantics and check all the details work.