Re: [MMUSIC] Partial Offer/Partial Answer draft

[Adam Roach <[e-mail]>]

On 10/15/13 12:53, Martin Thomson wrote:
> On 15 October 2013 10:05, Adam Roach <[e-mail]> wrote:
>> True, but this has its own point of brittleness: if a media section is part
>> of two groups of the same type, you're back to the issue of not being able
>> to clearly indicate which of the groups this rgroup attribute applies to.
> This specifically avoids that brittleness - the identifier in the
> a=rgroup: is an identifier for the *group*, so if a media section is
> part of two groups of the same type, then it's going to get two
> different group identifiers.
>
> m=whatever
> a=mid:blah
> a=rgroup:LS groupA
> a=rgroup:LS groupB
>
> Rather than identify one member of a group in order to identify the
> group by proxy, it identifies the group directly.  The cost is that
> you can't use 5888 any more.

Oh, I see. You're throwing 5888 out the window. I specifically didn't 
want to do that. The approaches I described were meant to be a means of 
communicating incremental changes to existing 5888 groups in a way that 
otherwise maintained 5888 processing as-is.

I think this is somewhere that backwards compatibility is actually 
somewhat important.

/a