[Dime] Review of draft-ietf-dime-rfc4005-bis

[Mark Jones <[e-mail]>]

(The issue tracker tool doesn't want seem to create account for me so
I'm posting here.)

>From the diff with rfc4005, I realize my comments relate to the
original text rather than Glen's changes but I think they're valid
candidates for the bis.

---
Section 3.9.  Accounting-Request (ACR) Command

"Either the Acct-Application-Id AVP or the Vendor-Specific-
Application-Id AVP MUST be present."

mj> I understand that the Acct-Application-Id is required for
backwards compatibility (it is still redundant since the same app id
is in the command header) but don't see why
Vendor-Specific-Application-Id would ever need be present. The same
comment applies to Section 3.10.

---
Section 4.1.1.  QoSFilterRule

"The QosFilterRule format is derived from the OctetString AVP Base
Format.  It uses the ASCII charset."

mj> RFC5777 defines Diameter AVPs that represent QoS and IP filter
rules and even includes a NASREQ example. I'd like to see the
ASCII-based variants deprecated but even if that idea doesn't fly, I
think it would be useful to mention RFC5777 here as offering an
alternative.

---
Section 4.2.5.  Called-Station-Id AVP

"It SHOULD only be present in authentication
and/or authorization requests."

mj> Why is this recommendation here? This AVP is commonly used in
RADIUS accounting requests. Same comment for Calling-Station-Id AVP in
Section 4.2.6.

---
Section 4.5.8.  Tunnel-Assignment-Id AVP

mj> s/should/SHOULD/g
mj> Last paragraph appears to be missing normative statements in the
first two sentences. Was this intentional?

Regards
Mark