Re: [radext] [IANA #935820] Protocol Action: 'RADIUS Extensions for IP Port Configuration and Reporting' to Proposed Standard (draft-ietf-radext-ip-port-radius-ext-16.txt)

[Alan DeKok <[e-mail]>]

On Nov 10, 2016, at 10:04 PM, Dean cheng <[e-mail]> wrote:
> 
>   The "type" field is 8-bit long so there are possibly 256
>   values. RFC6929 does not say the value "0" is "Unassigned"
>   or "Reserved". In the Radius registry, the value "0" for
>   attribute 241.1 is "Reserved". For consistency, I'd suggest
>   to follow the same. The maximum value is 256. 

  https://tools.ietf.org/html/rfc6929#section-10

   Allocation of an Attribute Type value "TYPE" using the new "Extended
   Type" format results in allocation of 255 new Attribute Type values
   of format "TYPE.1" through "TYPE.255".

i.e. zero is not for allocation.

>   Note each TLV in Section 7.3 may have more than one parent
>   Radius attributes (e.g., 241.TBD1.1 and 241.TBD2.1). Therefore
>   I'm not sure if these TLVs should go to Radius registry
>   (Section 7.3 only requests for "allocation" for TLV types).
> 
>   If need to define these TLV types in the registry, it must
>   say that these TLVs are independent and may associate with
>   more than one attributes; but I don’t see any example like
>   this in current registry.  
> 
>   Alan - what is your opinion?

  This is a new thing to RADIUS.

  I suggest creating a new RADIUS TLV registry:

name: RADIUS IP Port Configuration and Reporting TLVs

contents: all of the TLVs which have more than one parent.

Then, each parent TLV can be marked as NOT having child attributes.  Instead, they should be marked as using the new registry for their children.

  Alan DeKok.