Re: [sipcore] AD Evaluation of draft-ietf-sipcore-content-id-05 - PULL REQUEST

[Christer Holmberg <[e-mail]>]

Hi,

I really think that mandating global uniqueness, for a possible use-case that nobody is still aware of, is overkill.

IF such use-case comes up, someone can always update the spec in order to mandate a globally unique value for such use-case. Alternatively, the value can be used together with some other value(s) (Call-ID, transaction-id, CSeq etc etc etc) in order to create a unique reference.

Regards,

Christer 

-----Original Message-----
From: Ben Campbell [mailto:[e-mail]] 
Sent: 31 May 2017 22:03
To: Christer Holmberg <[e-mail]>
Cc: A. Jean Mahoney <[e-mail]>; [e-mail]; [e-mail]
Subject: Re: AD Evaluation of draft-ietf-sipcore-content-id-05 - PULL REQUEST


> On May 28, 2017, at 1:40 PM, Christer Holmberg <[e-mail]> wrote:
> 
> Hi,
> 
>>>>> I have created a pull request, based on your comments:
>>>>> https://github.com/cdh4u/draft-content-id/pull/6
>>>> The diff looks fine. We probably want to make sure the WG shares 
>>>> the opinion that the Content-ID will never be referenced from 
>>>> outside the SIP message.
>>>> Jean, do  you have thoughts on that from the shepherd perspective?
>>> 
>>> The WG did discuss whether the Content-ID could be used outside of the message. 
>>> The takeaway was, that since a SIP header has non-MIME fields, the 
>>> Content-ID can't really refer to the entire message, and thus would not be useful outside the message.
>> 
>> There seems to be two ideas intertwined there; namely the idea of 
>> what a content-ID identifies, and the idea of whether a content-ID could be referred to from outside the containing SIP message.
>> But I take your comment to mean that both were discussed. Is that correct?
> 
> As far as I remember, we did not discuss the possibility of referencing a body outside the SIP message.
> 
> However, nobody has requested for that possibility, so I don't think we need to cover it. If someone, as some point, see a need for it, he/she can update the spec and define how it is done, what impacts it has on the uniqueness etc.

It’s kind of hard to define extra-message references later if the initial version does not require global uniqueness. If people want to leave that option open, then global uniqueness may still make sense now. I’m okay with it either way as long as the WG has thought it through, and hasn’t just picked it arbitrarily.

> 
> Regards,
> 
> Christer
> 
> 
> 
>>>> However, I wasnąt sure how to address the following comment:
>>>> "1.2 and 1.3: A sentence or two that more strongly contrasts "body 
>>>> part" vs "message-body" would be helpful. I think that some people 
>>>> will think of a message-body as still a body-part.˛ I think section
>>>> 1.1 describes the difference between a message-body and a body-part. 
>>>> I donąt think we should copy/paste that in sections 1.2 and 1.3. 
>>>> Or, did I misunderstand you comment?
>>> On reflection, I think this might be fine like it is. I know that 
>>> some people casually refer to the entire body as still a “part”, but 
>>> that doesn’t seem to be reflected in the MIME RFCs. Let’s see if 
>>> anyone comments in LC.
>>>> Regards,
>>>> Christer
>