Re: [RAI] Global Service Provider ID draft-pfautz-service-provider-identifier-urn-01

[Hadriel Kaplan <[e-mail]>]

I think we all agree this is important - no one's trying to say "go away".  And you're right it will probably require an informational RFC no matter what. 
Other comments inline...

On Sep 30, 2011, at 11:41 AM, Richard Shockey wrote:
> There is a preference here for a new fixed length identifier in a non
> polluted name space so the counter argument is why is there a problem here?

As I said at the Quebec meeting, as far as I can tell the "fixed-length" aspect is the encoding model of the number on-the-wire, not the namespace value.  I.e., IANA assigns AT&T the number 1234, and it just happens that we decide to encode the number in SIP/ENUM/DRINKS message fields padded out to be the string "0000001234", if that's what's needed.  Whereas if this SPID value ends up in a new AVP in DIAMETER (and I'm pretty sure it will), it could be encoded as a 4-byte Unsigned32 data type rather than a 10-byte OctetString or UTF8String.

Is that correct?

We need to understand whether that's correct or not, or if the IANA registry literally needs to be a registry of fixed-length digit-strings instead.


> We rejected btw the IANA Enterprise namespace as well for similar reasons.

Enterprise numbers wouldn't be right for other reasons anyway.  For example, although SNMP never really spelled out what a PEN means, IANA usually only likes to give one PEN per company; whereas ITADs were never meant to have that restriction or meaning.  They're an administrative domain, so one company can have many. 

-hadriel