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

[Harald Alvestrand <[e-mail]>]

On 10/19/2012 04:26 PM, Matthew Kaufman wrote:
> Iñaki Baz Castillo [[e-mail]]:
>> SIP and XMPP are signaling protocols using SDP and fit very well into
>> the state machine defined in RFC 3264. In WebRTC we DON'T have a
>> (network) signaling protocol and, thus, it's the JavaScript
>> responsability to respect or not RFC 3264 rules, but those rules
>> cannot be mandated at the media stack level.
>>
>> So honestly I don't fully understand what we are discussing in this
>> thread. Regards.
Inaki is not alone.
The text in the JSEP draft could definitely bear improving. But I have 
problems parsing the definition of "compliant" here:
> What I am hoping to discuss is whether the JSEP API (which repeatedly claims to be exposing an RFC3264 SDP O/A compliant API surface*) has or has not "relaxed" RFC3264. I think it is pretty clear that it has (in that it does not in fact enforce any of "MUST NOT" called out in RFC3264 at the API level),
in that I would expect a "compliant" API to be one that permits an RFC 
3264 protocol machine to be built on top of it, which doesn't make the 
sentence above logical
>   and so now that we don't have an API that complies with SDP O/A
.... a premise I don't believe.
>   *any* API that doesn't comply with SDP O/A should be equally acceptable.
.... but this conclusion is still completely bizarre to me.

RFC 3264 compliance (for whatever definition of compliance) is just one 
aspect of the considerations of what main direction of API to choose. 
Anyone who wasn't part of the WEBRTC discussions is free to consult the 
archives of the recent discussion for the other reasons put forward.

"Equally acceptable" is a very strange phrase to use in this discussion 
(which, anyway, is in the wrong forum).