Re: [IPsec] Draft-ietf-ipsecme-ipv6-ipv4-codes

[Paul Wouters <[e-mail]>]

On Mon, 6 May 2019, Valery Smyslov wrote:

>>> That would make sending these notifies dependent on the content of request.
>>> So, the tradeoff is whether saving eight bytes justifies complication of state machine.
>>
>> I wouldn't call that complicated the state machine. You are not adding
>> new states or transitions, and you already keep a list of received
>> payloads for this state/exchange I hope :P
>
> True, I wasn't precise enough. The complication is that in the current approach
> the responder sends these notifications blindly, sending them doesn't depend
> on the content of CP request.

RFC 7296 found that argument weak :)

    Note that no recommendations are made in this document as to how an
    implementation actually figures out what information to send in a
    response.  That is, we do not recommend any specific method of an
    IRAS determining which DNS server should be returned to a requesting
    IRAC.

For example, you can always reply with INTERNAL_IP4_DNS and INTERNAL_IP4_NBNS
if the client is requesting only an INTERNAL_IP6_ADDRESS and either
these first two CFG payloads or not. But I hope you added "complicated code"
to not send those :)

sub-optimal clients get sub-optimal answers.

Paul