Re: [MMUSIC] [rtcweb] Updating JSEP and BUNDLE

[Christer Holmberg <[e-mail]>]

Hi,

...

>>>>>>> Secondly, I can definitely live with the work being done in a revivied
>>>>>>> specifically charted RTCWeb WG. However, SDP Offer/Answer is used in several
>>>>>>> places for different usages. Something which was considred very much when
>>>>>>> developing BUNDLE. Thus, I think it is relevant that RTCWeb will keep MMUSIC
>>>>>>> informed about their work. So that RTCWeb doesn't end up missing an important
>>>>>>> aspect of this, a run into issue when forming the IETF consensus needed to
>>>>>>> publish this document. 
>>>>>>
>>>>>> I think the revised charter takes this into account.  Please let me know if it needs any other corrections.
>>>>> 
>>>>> It needs to be clear that the charter is about fixing the misalignment between the BUNDLE and JSEP specifications - nothing more, nothing less.
>>>>>
>>>>> As noted in the spec blurb, we also need to consider the existing deployments and how any changes at this point may affect them. Specs that implementors can't implement are not useful. 
>>>>
>>>> Absolutely. When we discuss how to solve the misalignment between the JSEP and BUNDLE specifications we need to take existing deployments into consideration.
>>>>
>>>> But, any issue that is not related to the misalignment between the JSEP and BUNDLE specifications shall be outside the scope.
>>>
>>> Agreed that we want to keep this narrowly focused. I believe the existing blurbs added to JSEP and BUNDLE outline the scope well, in that we are trying to solve a specific problem, but the solution space for said
>>> problem may need to consider options beyond what JSEP and BUNDLE currently specify. 
>>>
>>> On that topic: to understand the importance of the BUNDLE v41 change regarding answers and subsequent offers, does anyone recall the technical reasons behind this change?
>>
>> The change was suggested by Taylor:
>>
>> https://mailarchive.ietf.org/arch/msg/mmusic/-jENmu4iaapy7NdJ9_SLHpNjSDc/
>>
>> Based on that text, there doesn't seem to be a technical reason, but rather just a wish to not duplicate unnecessary information (the port number).
>>
>> I know you have been trying to reach Taylor, so hopefully he can confirm whether there is a technical reason or not :)
>
> I haven't heard from him yet, but my read of that thread is that the change is entirely focused on cleaner syntax. 

So, assuming that is true, from a pure technical perspective I see 3 different options:

1) Keep the BUNDLE way
2) Keep the JSEP way
3) Allow both ways (at the end of the day, once the endpoints have negotiated usage of BUNDLE, only the port of the tagged m- line really matters).

Regards,

Christer