Re: [MMUSIC] unified plan glareless atomicity improvement

I am not sure if this proposal answer this topic .. I had a similar
proposal with uni-directional handling w.r.t glare ...

http://tools.ietf.org/id/draft-nandakumar-rtcweb-glare-handling-00.txt

Cheers
./S
PS: Its been a while i read the above draft myself .. So please forgive me
, if this is totally irrelevant :)

On Wed, Jul 31, 2013 at 2:16 PM, Martin Thomson <martin.thomson@gmail.com>wrote:

> The draft currently doesn't say much about how the glareless
> negotiation features relate to a=sendrecv/sendonly/etc...  I think
> that using recvonly and sendonly could reduce the probability of
> encountering glare.
>
> It's a detail.  And obviously, there are tradeoffs here.  By choosing
> to use sendonly in an offer, you foreclose the ability of the answerer
> to use that m-line for their streams.  So you would need to generate
> extra recvonly m-lines in order to enable the 1-RTT negotiation.
> There may be other implications that then stem from using extra
> m-lines, especially in legacy interop cases, e.g. where BUNDLE
> negotiation fails, or you can't access SSRCs etc..., each of which
> have various negative consequences.
>
> I think that the only guidance needed is something like:
> "Implementations MAY choose to exclusively use m-lines with a=sendonly
> or a=recvonly in order to further reduce the probably of encountering
> glare."
> _______________________________________________
> mmusic mailing list
> mmusic@ietf.org
> https://www.ietf.org/mailman/listinfo/mmusic
>