Re: [DNA] MUST NOT add DNA Option when landmark RA = NO?(draft-pentland-dna-protocol-00.txt)

[Greg Daley <greg.daley@eng.monash.edu.au>]

Hi Erik,

Erik Nordmark wrote:
> Greg Daley wrote:
> 
>> Can't the host use the prefixes learnt from the DNAO to not be
>> surprised by prefixes in responding RAs (which may not be DNA RAs)
>> to its existing solicitation?
> 
> 
> What did you have in mind the host would actually do to be "less 
> surprised"?

I'm still cooking something, but to talk now on list
wouldn't really help the discussion

Please contact me if it's important to know now.

> If it records the DNAO prefix somehow, what would it then do when it
>  - receives an RA with a prefix which was recorded from the DNAO

It would be able to tell they belong to the same link.

>  - receives an RA with a new prefix which was not recorded from a DNAO?

That depends...

If it was 'surprised' I guess it would be treated as a hint of change
(like in CPL).

I think it's possible to do better than CPL with the learning phase
though (since you have an explicit CPL in the CompleteRA).

> FWIW the conceptual data structures in RFC 2461 can only capture 
> prefixes for which all the information in the PIO is present, i.e. the 
> lifetimes and the flags must be includes.

Yes, that's right.   It's not necessary to use exactly these
data structures, though if you're not using them for address
configuration, just for link identification.

(Though the link identification logic could be assisted if you
knew them).

Greg