Re: [MMUSIC] Simulcast worries: Fallback in the a=rid language

[Adam Roach <[e-mail]>]

On 10/22/15 1:13 PM, Harald Alvestrand wrote:
> Den 22. okt. 2015 16:43, skrev Adam Roach:
>> On 10/22/15 00:37, Harald Alvestrand wrote:
>>> If I tell someone I can't take more than X from them (recv), it's
>>> dangerous if they ignore it, because they can overrun my limitations
>>> through ignorance.
>> And if they don't support the simulcast mechanism at all, they'll do the
>> same thing. You're asking us to solve a problem that is unrelated to the
>> mechanism.
> I don't understand this comment.
>
> If that (telling the other guy not to send more in a simulcast than you
> want to handle) isn't the problem the mechanism was supposed to solve,
> what IS the point of the mechanism?
>
> I'm assuming that the correspondent that doesn't understand simulcast is
> restricted by other attributes like a=imageattr (well, sort of), a=fmtp
> and b=, so that this is regarded as a problem this mechanism doesn't
> have to solve.
>

Right, and all of those still apply, even when they *do* understand 
simulcast. 
https://tools.ietf.org/html/draft-pthatcher-mmusic-rid-02#section-9


In any case, this is all kind of academic, unless you're proposing that 
we change things to work in the way you assert is broken. The current 
situation (in the spec) is that those RIDs containing restrictions that 
you don't understand are ignored. Not the restriction, the entire RID. I 
think your point that we might be able to allow the RIDs to say around 
for streams you're receiving (it doesn't matter if the far end restricts 
the stream it's sending in a way you don't understand) is a good one 
that we should probably take into account.

/a