Re: [MMUSIC] Fwd: New Version Notification for draft-ietf-mmusic-msid-01.txt

[Martin Thomson <[e-mail]>]

On 15 August 2013 15:58, Harald Alvestrand <[e-mail]> wrote:
> I wondered about that too.
> We have two IDs to convey:
>
> - The ID of the MediaStream. The grouping can be a new semantic for the
> grouping framework, but the actual ID value has to be carried somewhere
> - it has to be associated with the group, so it can't be a session-level
> attribute.
>
> - The ID of the MediaStreamTrack. Now that MediaStreamTracks correspond
> to m-lines, it makes sense to carry this in a media-level attribute.
>
> I'm sure we can design another scheme for carrying those IDs around.
> But I do wonder what the positive value of restructuring this is.

The MediaStream identifier gets lost if the grouping framework is
used.  Unless you define some sort of new mapping attribute, which
could end up looking like a=msid if you aren't careful.

The MediaStreamTrack identifier could be doubled onto the a=mid value.

I've always been lukewarm on transmitting identifiers in this fashion,
but I find it hard to find a better (i.e., cleaner, more obvious, less
redundant) solution than msid, except perhaps this one:

a=appgroup:MediaStreamId Track1 Track2
a=group:LS Track1 Track2
m=...
a=mid:Track1
m=...
a=mid:Track2

--Martin