Re: DRAFT Montreal minutes - X#NodeArchitecture comments

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Aug 17, 2006, at 6:43 AM, Paul Koning wrote:

>  Mallikarjun> Hi Ken, Can you please point me to RFC 3720 text that
>  Mallikarjun> says NotUnderstood is always an error?
>
>  Mallikarjun> The closest I can find is in section 5.2, page 55:
>
> Page 54, second paragraph:
>
>    The constants "None", "Reject", "Irrelevant", and  
> "NotUnderstood" are
>    reserved and MUST ONLY be used as described here.  Violation of  
> this
>    rule is a protocol error (in particular the use of "Reject",
>    "Irrelevant", and "NotUnderstood" as proposed values).
>
> "as described here" is as responses to negotiating keys.  The text
> already explicitly disallows their use as proposals.
>
> A declaration isn't exactly the same as a proposal, but it certainly
> isn't a reply to a proposal, so by the above text "NotUnderstood" is
> prohibited.
>
> The 3720 bug is this: an implementation that doesn't understand a key
> doesn't know whether that key is declarative or negotiated.  If it
> assumes the latter it would have to reply NotUnderstood.  If in fact
> it was a declarative key then by the current text that's a protocol
> violation.
>
> A good way to fix this is to add two rules:
>
> 1. If a key is not understood, it is assumed to be a negotiation
>    key (and a NotUnderstood reply must be sent).
> 2. If a declarative key is sent, and a reply is received from the
>    other end with that key and a value of NotUnderstood, this is
>    a protocol violation only if the key is required to be implemented.
>    If it was not required (i.e., X# keys) then a NotUnderstood
>    reply must be silently ignored.

I like the proposed fix. The two comments I have are:

1) While X# keys are not required to be understood, any future  
"normal" keys we define will also fall into this category for  
existing devices.

2) I note that a device may chose to pretend it doesn't understand a  
key that isn't mandatory, even though it really does. For  
X#NodeArchitecture, a device may answer with differing levels of  
detail. The lowest level may well be to pretend that the key's not  
understood. In such a case, the device MUST fully pretend it doesn't  
understand, and as per rule (1) above, it MUST not offer the key and  
MUST respond with "NotUnderstood."

Note: this is a different behavior from understanding  
X#NodeArchitecture and choosing not to declare a value to the other  
side.

Take care,

Bill
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFE5LpfDJT2Egh26K0RApccAJ9gvQ1wVdAD/U66DkSMUXM8hbLgYACfTXMX
4O7tC0z/Uatk/LfNJP/9+SY=
=uObj
-----END PGP SIGNATURE-----

_______________________________________________
Ips mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/ips