Re: [Gen-art] Gen-art LC review of draft-saintandre-impp-call-info-02

[Adam Roach <[e-mail]>]

On 4/24/13 09:19, Peter Saint-Andre wrote:
> Hi Elwyn, thanks for the review.
>
> On Apr 23, 2013, at 2:12 PM, Elwyn Davies wrote:
>> Generic comment about SIP Header Field Parameters registry: For the 
>> uninitiated this registry is rather opaque.  Some parameters, such as 
>> the Call-Info purpose parameter for which an extra value is defined 
>> here, have predefined values.  However the predefined values 
>> themselves are not in the registry and just giving a whole RFC 
>> reference for places where values are defined is not very helpful. 
>> For example, in the case of Call-Info, the initial predefined values 
>> of purpose are buried in the Call-Info rule in the ABNF in Section 
>> 25.1 of RFC 3261;  also, Section 20.9 describes the predefined values 
>> (such as "icon") as 'parameters' rather than values of 'purpose'.  It 
>> would probably be helpful to either improve the references in the 
>> registry table or actaully quote the possible predefined values in 
>> the table.
>>
>
> I agree, but as you say that's an issue with the SIP Header Field 
> Parameters registry in general. I started to go down the path of 
> fixing the registry as a whole, but I think I'd rather leave that for 
> 3261bis to tackle (sometime before the heat death of the universe).

Yes, the registry is a mess, largely due to the rather whimsical 
preferences -- which would vary from year to year -- of the SIP 
community regarding which fields needed a registry and which did not. 
It's grown to be the way it is somewhat organically, and I agree that 
the overall SIP registry is something of a mess.

I'm not sure we need to wait for a 3261bis to fix things, as I doubt we 
could find someone with the fortitude to take on an effort that large. 
It might be worth finding someone to work on a "registry overhaul" 
document that attempts to make the whole thing more coherent. I'll be 
keeping this in mind as a potential SIPCORE item.

/a