[MMUSIC] The role of signaling

[Martin Thomson <[e-mail]>]

I actually think that noplan bears serious consideration, but only
with respect to what it says about the role of signaling.

Plan A/B assume some things about these questions that we clearly
haven't reached consensus on.  The questions that I think we've been
working towards are:

 - What role does signaling play in session establishment?
 - What information does signaling convey vs. what information is
conveyed in the signaled medium?
 - Does SDP represent ALL signaling?  (Clearly not.)  What parts of
signaling is SDP responsible for?

I don't think that we resolve anything on A/B until we agree on what
SDP is and does.  Of course, once we agree on principles, we are
always free to choose expediency over those principles.

I think that Emil was right to recognize that SDP O/A doesn't need to
be the only signaling mechanism.  The fact that WebRTC is using a
signaling protocol as part of an API surface is, in my opinion, the
reason for this affliction.  It appears as though SDP is the carpet to
sweep all the muck under when it doesn't fit anywhere else, simply
because we took the other options (read: SIP) away.  That puts a lot
of pressure on SDP.

For instance, to pick an easy example, adding identity assertions to
SDP could be the wrong choice.  In SIP, identity is the domain of the
SIP part of signaling.