Re: [sipcore] [Technical Errata Reported] RFC3261 (4553)

[Paul Kyzivat <[e-mail]>]

On 12/14/15 8:31 PM, Ben Campbell wrote:
> On 14 Dec 2015, at 18:52, Paul Kyzivat wrote:
>
>> On 12/14/15 6:21 PM, Ben Campbell wrote:
>>> On 14 Dec 2015, at 17:00, Paul Kyzivat wrote:
>>>
>>>> On 12/14/15 5:29 PM, Ben Campbell wrote:
>>>>
>>>>> In any case, it seems to me from discussion so far that we do not
>>>>> have a
>>>>> settled consensus on what the default value of the handling parameter
>>>>> should be when C-D is not present. That suggests this should be
>>>>> done in
>>>>> an I-D, not in an erratum.  If people disagree with that, please
>>>>> speak up.
>>>>
>>>> I don't know if we have consensus. But is there any plausible argument
>>>> for why the default should be different when the C-D is missing from
>>>> when it is present but doesn't have a handling parameter?
>>>>
>>>
>>> I agree that this is already settled for the case where C-D is present
>>> but with no handling parameter, and consistency is a strong argument.
>>> But do we think it makes sense to say "If the content is critical, put
>>> in C-D" rather than "If the content is _not_ critical, put in C-D"? Do
>>> we really think "required" is or will be the most common case, or that
>>> the risk of accidentally failing to render critical content is high
>>> enough to default to failure?
>>>
>>>
>>> I'm personally on the fence here, if everyone paying attention thinks
>>> consistency with the default when C-D is present is the correct way to
>>> handle this, and was the original intent in 3261, then an erratum makes
>>> sense. But if we feel the need to discuss this further, this probably
>>> should come in the form of an I-D, even if the final answer is the same.
>>
>> Well, in the case where there is no C-D, and the C-T is SDP, then the
>> default is handling=required. Is that enough to convince you?
>
> SDP has a rather privileged place in SIP messages.

Yes, it does. But some effort was made to arrange a consistent set of 
general rules for bodies that is backward compatible with that.

> But let me turn that around. Are you convinced? Does anyone else thing
> this needs further discussion? If you are, and no one else speaks out
> really soon now[1], I would approve an erratum without the SHOULD to
> MUST bit.
>
> [*] Meaning, "before such an erratum arrives in my email."

Yes, I'm convinced.

	Thanks,
	Paul