Re: [siprec] What does an MS object represent? Can we pick one of theseoptions?

Charles,

I think it was Paul who first suggested that a=content might not be rich enough, and I was just discussing the consequences. So perhaps Paul should respond.

John

> -----Original Message-----
> From: Charles Eckel (eckelcu) [mailto:eckelcu@cisco.com] 
> Sent: 10 June 2011 18:36
> To: Elwell, John; Paul Kyzivat (pkyzivat); Ram Mohan R (rmohanr)
> Cc: siprec@ietf.org
> Subject: RE: [siprec] What does an MS object represent? Can 
> we pick one of theseoptions?
> 
> Hi John,
> 
> Given that the syntax for the content attribute in RFC 4796 allows for
> the assignment of additional extension values:
> 
>        content-attribute   = "a=content:" mediacnt-tag
>        mediacnt-tag        = mediacnt *("," mediacnt)
>        mediacnt            = "slides" / "speaker" / "sl" / "main"
>                              / "alt" / mediacnt-ext
>        mediacnt-ext        = token
> 
> do you still think it is not rich enough? It seems to be that 
> given the
> content and label attributes plus the grouping framework, we 
> can express
> pretty complex and detailed relationships. That said, it may 
> be simpler
> to express in metadata than in SDP. Is that the motivation?
> 
> Cheers,
> Charles 
> 
> > -----Original Message-----
> > From: siprec-bounces@ietf.org [mailto:siprec-bounces@ietf.org] On
> Behalf Of Elwell, John
> > Sent: Tuesday, May 24, 2011 11:15 PM
> > To: Paul Kyzivat (pkyzivat); Ram Mohan R (rmohanr)
> > Cc: siprec@ietf.org
> > Subject: Re: [siprec] What does an MS object represent? Can we pick
> one of theseoptions?
> > 
> > Given that SDP's a=content might not be rich enough, I think we need
> to define something new for
> > metadata, independent of a=content. The SRC could choose to 
> derive the
> metadata content attribute from
> > the a=content attributes of the contributing CS streams (that is an
> SRC decision), but should not use
> > the a=content label in the RS SDP.
> > 
> > John (as individual)
> > 
> > > -----Original Message-----
> > > From: Paul Kyzivat [mailto:pkyzivat@cisco.com]
> > > Sent: 24 May 2011 18:07
> > > To: Ram Mohan R (rmohanr)
> > > Cc: Elwell, John; siprec@ietf.org
> > > Subject: Re: [siprec] What does an MS object represent? Can
> > > we pick one of theseoptions?
> > >
> > >
> > >
> > > On 5/24/2011 12:52 PM, Ram Mohan R (rmohanr) wrote:
> > > > Paul/ John,
> > > >
> > > > I am ok to add attribute "content" in MS block of the model.
> > > >
> > > > The question is do we want to map this attribute in the model to
> > > > 1) The Session Description Protocol (SDP) Content Attribute
> > > defined in
> > > > RFC 4796 OR
> > > > 2)define our own content in the metadata XML which can take
> > > values as
> > > > "audio" / "main video" / "presentation video" / "current
> > > speaker video"
> > > > as suggested by John.
> > >
> > > Well...
> > >
> > > ISTM that if there is already an a=content in the CS SDP,
> > > then we would
> > > want to report that in the metadata.
> > >
> > > If there is a pair (one from offer, one from answer, same
> > > m-line offset)
> > > and there is *not* an a=content that pairs them, then we
> > > would want to
> > > report it anyway.
> > >
> > > But if we do that we have a problem with names and conflicts.
> rfc4796
> > > doesn't define a lot of names, nor does it say whether 
> corresponding
> > > streams in offer & answer must have the same names.
> > >
> > > So I think we have something similar but not quite the same,
> > > and we need
> > > to define something new. It must permit identifying streams
> > > even though
> > > we know nothing of their purpose.
> > >
> > > 	Thanks,
> > > 	Paul
> > >
> > _______________________________________________
> > siprec mailing list
> > siprec@ietf.org
> > https://www.ietf.org/mailman/listinfo/siprec
> 