Re: [MMUSIC] "New" style for defining SDP attributes

Christer,

Note that I changed the subject line here. My intent was to start a 
different discussion that isn't specifically related to trickle-ice-sip. 
The intent instead is to discuss if we should change 4566bis to be more 
explicit about how to define attributes.

	Thanks,
	Paul

On 5/13/18 4:47 AM, Christer Holmberg wrote:
> Hi,
> 
> My suggestion would be to simply reference the syntax. Something like:
> 
> trickle-ice-sdpfrag =   session-level-fields
>                                        pseudo-media-descriptions
> session-level-fields = bundle-group-attribute CRLF          \
>                                        ice-lite-attribute CRLF                      \
>                                        ice-pwd-attribute CRLF                     \
>                                        ice-ufrag-attribute CRLF                    \
>                                        ice-options-attribute CRLF                \
>                                        ice-pacing-attribute CRLF                  \
>                                        end-of-candidates-attribute CRLF   \
>                                        extension-attribute-fields
> 
> The attribute fields follow the generic syntax for SDP attributes [RFCXXXX]. The list below reference the specific syntax for each attribute field.
> 
> ice-lite-attribute              ; as ice-lite defined in [RFCXXXX]
> ice-pwd-attribute            ...
> ice-ufrag-attribute           ...
> ice-pacing-attribute         ...
> ice-options-attribute       ....
> 
> Then we don't need to worry about "new" and "old" in this draft. We simply define the high-level syntax for the sdpfrag message body, and reference the appropriate specifications for the attributes etc.
> 
> In addition, I think it would be useful to say that in order to use an extension attribute, there needs to be a specification for how that attribute is used within a sdpfrag message body. Otherwise someone could include any SDP attribute just like that.
> 
> Regards,
> 
> Christer
> 
> 
> -----Original Message-----
> From: mmusic [mailto:mmusic-bounces@ietf.org] On Behalf Of Paul Kyzivat
> Sent: 12 May 2018 19:41
> To: mmusic@ietf.org
> Subject: [MMUSIC] "New" style for defining SDP attributes
> 
> On 5/10/18 12:59 PM, Christer Holmberg wrote:
> 
>> That's a bad example since a=bundle-only doesn't have a value.
>>
>> A better example is the tls-id attribute, defined in Section 4 of
>> https://www.ietf.org/id/draft-ietf-mmusic-dtls-sdp-32.txt.
>>
>>
>> I'd rather prefer ( just personal preference ) the style of
>> https://tools.ietf.org/html/draft-ietf-mmusic-rid-14, where the
>> registration in section 11 refers to the grammar in section 10.
>>
>> The ABNF is not according to the new style.
>>
>> And, just to clarify, my comments are not based on my personal
>> preference, but on my understanding of how SDP attributes are to be
>> defined nowadays.
> 
> I've been finding that the "new" style that I have been trying to get institutionalized isn't well enough specified. I'm thinking that we need to firm it up, and preferably get that written down in 4566bis.
> 
> 	Thanks,
> 	Paul
> 
> _______________________________________________
> mmusic mailing list
> mmusic@ietf.org
> https://www.ietf.org/mailman/listinfo/mmusic
> 