Re: [MMUSIC] BUNDLE DECISION Q8: SDP 'bundle-only' attribute - Suggested 3264 update

[Christer Holmberg <[e-mail]>]

Hi,

So, let's do things in steps:

First, does anyone object that we need to, SOMEHOW (we'll discuss the actual text etc later), update RFC 3264?

Regards,

Christer

-----Original Message-----
From: Paul Kyzivat [mailto:[e-mail]] 
Sent: 18. marraskuuta 2013 20:08
To: Christer Holmberg; Stach, Thomas; Ravindran, Parthasarathi (NSN - IN/Bangalore); Martin Thomson; Adam Roach
Cc: [e-mail]
Subject: Re: [MMUSIC] BUNDLE DECISION Q8: SDP 'bundle-only' attribute - Suggested 3264 update

On 11/18/13 9:19 AM, Christer Holmberg wrote:
> Hi,
>
> The issue is not whether a 3264 implementation will break or not.
>
> The issue is more administrative, and how we are allowed to write 
> standards in IETF.
>
> Currently, 3264 says:
>
> 1) Port zero in an Offer means that the stream must not be used; and
>
> 2) An Offer with port zero must be answered with port zero
>
> As far as I know, we can not simply override that by defining an SDP 
> attribute ('bundle-only'). Unless we can move inside the boundaries 
> set by 3264, we need to update 3264.
>
> Please correct me if I'm wrong, though. I guess some chair/AD input 
> would be useful here.

I queried Adam for his opinion so we could have unanimity among chairs, but I haven't heard back from him.

I agree with Martin, that we need to formally declare that we are updating 3264, but that we don't have to provide an explicit delta to
3264 to do so.

> IN ANY CASE, I really hope we are not going to spend too much time on 
> this issue - we have more important things to do.

Agree.

	Thanks,
	Paul

> Regards,
>
> Christer
>
> ----------------------------------------------------------------------
> --
> *From:* Stach, Thomas [[e-mail]]
> *Sent:* Monday, 18 November 2013 6:45 PM
> *To:* Ravindran, Parthasarathi (NSN - IN/Bangalore); Martin Thomson; 
> Christer Holmberg
> *Cc:* [e-mail]; Paul Kyzivat
> *Subject:* RE: [MMUSIC] BUNDLE DECISION Q8: SDP 'bundle-only' 
> attribute
> - Suggested 3264 update
>
>>
>> Hi Thomas,
>>
>> <snip> An answerer compliant with 3264 and not supporting bundle 
>> would ignore a=bundle-only and would set the port=0 in its answer. 
>> </snip>
>>
>> The above statement clearly shows how RFC 3264 implementation will 
>> break with "bundle-only" attribute proposal.
> [TS] I'm not sure what 'break' should mean here. If it means "reject 
> the m-line by setting port= in the answer", that is the intended behavior.
> My statement was based on
> http://tools.ietf.org/html/draft-roach-mmusic-unified-plan-00#section-
> 3.1 Here, the a=bundle-only attribute is combined with port=0 in the 
> offer.
> The port=0 in the offer should guarantee that an answerer does not use 
> the m-line.
>> RFC 3264 would have
>> properly negotiated in case the port is allocated in the initial 
>> offer instead of port 0.
> [TS] That is not what a=bundle-only wants to achieve.
> The semantics of using a valid port in the offer together with 
> a=bundle-only are not clear to me.
> I would regard that as an invalid combination.
>
> Regards
> Thomas
>>
>> Thanks
>> Partha
>>
>> -----Original Message-----
>> From: [e-mail] [mailto:[e-mail]] On 
>> Behalf Of ext Stach, Thomas
>> Sent: Monday, November 18, 2013 4:54 PM
>> To: Martin Thomson; Christer Holmberg
>> Cc: [e-mail]; Paul Kyzivat
>> Subject: Re: [MMUSIC] BUNDLE DECISION Q8: SDP 'bundle-only' attribute 
>> - Suggested 3264 update
>>
>> Inline
>>
>> > On 14 November 2013 10:28, Christer Holmberg 
>> > <[e-mail]> wrote:
>> > > I scanned through 3264, in order to find the sections that need 
>> > > to
>> be
>> > updated.
>> >
>> >
>> > A direct replacement doesn't seem necessary.  We aren't issuing a 
>> > new 3264.  A more direct explanation of what 3264 permits or 
>> > prohibits
>> and
>> > why this is being changed is probably more important.
>> >
>> > Maybe it's as simple as saying something like:
>> >
>> > RFC 3264 states that m-lines with zero ports are negotiated off.
>> This
>> > document updates RFC 3264 to permit the use of a zero port on 
>> > m-lines in combination with the a=bundle-only attribute.
>> [TS]  I agree with Martin.
>> An answerer compliant with 3264 and not supporting bundle would 
>> ignore a=bundle-only and would set the port=0 in its answer.
>> I don't see an issue with an offerer compliant with 3264 and not 
>> supporting bundle, since it wouldn't have included a=bundle-only (and 
>> wouldn't include a larger number of m-lines with port=0). The 
>> answerer supporting bundle would have to set port=0 in the answer to 
>> such an offer in order to comply with 3264.
>> In my opinion there isn't any change to the procedures in 3264 for an 
>> older implementation, so this isn't an update to 3264.
>> However, explicit text in bundle that describes this behavior would 
>> be justified.
>>
>>
>> Regards
>> Thomas
>>
>> > _______________________________________________
>> > mmusic mailing list
>> > [e-mail]
>> >https://www.ietf.org/mailman/listinfo/mmusic
>> _______________________________________________
>> mmusic mailing list
>> [e-mail]
>>https://www.ietf.org/mailman/listinfo/mmusic