[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: WGLC: Name Server Identifier Option





Dear colleagues,

There is consensus to publish this draft on the standards track, there have been sufficient reviewers, but there seems to be one issue that has not been completely resolved and that ended in a discussion between Rob and Peter.

This is what Rob proposed:
If the WG would prefer that we just use an empty NSID option rather
than the SI flag bit, that's a simple change.

Absent a specification for what non-empty NSID payload from client to
server would mean, I think it should be empty (name server MUST ignore
NSID payload, client MUST NOT/SHOULD NOT send NSID payload).  If and
when the WG specifies what client->server payload looks like, how it's
supposed to work, etc, that specification can just update this one if
the WG concludes that reusing the NSID option is appropriate.  I still
suspect that the client->server case is different enough that it
should be a separate option, but hey, I could be wrong.


The issue is whether to drop the SI flag and instead use the empty NSID OPTION as a signaling mechanism and MUST language for the server ignoring client NSID payload (as argued by Peter on the 28th of October).

I propose that if nobody screams we declare that the consensus position is that we drop SI and use empty NSID, and specify MUST, publish 01, and push it to the IESG without a new last call. Scream within a couple of days if you do not agree with this declaration of consensus :-)...

--Olaf
  Co-Chair.


-----------------------------------------------------------
Olaf M. Kolkman
NLnet Labs
http://www.nlnetlabs.nl/



Attachment: PGP.sig
Description: This is a digitally signed message part