Dnssd discussion ran long so I couldn’t make it, so will use email.
Here’s the summary…

Slides summarizing are at 
http://www.ietf.org/proceedings/87/slides/slides-87-pcp-11.pdf
Some things we did:

1)      get agreement on the requirements

2)      get agreement on what possible solutions would meet the requirements 
(we concluded that there were two, with support roughly split between them and 
potentially contentiously so)

3)      Appropriately phrase the questions to ask for consensus on (see slide 3)

4)      Do the consensus call on the list.

5)      We chose to use draft-resnick-on-consensus as our handbook for judging 
consensus.
One wrinkle we ran into was how to handle the few responses that weren’t public 
(i.e. sent just to the chairs).  As chairs, we shared all such responses with 
each other and with our AD, and ensured that such responses included no 
technical issues that weren’t brought up on the mailing list.  If there were, 
we made sure that such issues were then discussed publically (in this case 
there were no such technical issues raised in private responses).  There *were* 
various non-technical business preferences stated, some of which indicated that 
one or more implementers would only implement one of the two proposals and not 
bother with the other one.   We had to decide how this affected “consensus”.   
What we did is described in the slides, but to save you the suspense it turned 
out that when judged by technical arguments alone, we declared consensus on the 
one that the aforementioned implementers were much more likely to implement, so 
dodged the proverbial bullet on that one… if it was the other way, the protocol 
would likely be declared a failure in the RFC 5218 sense.

6)      We used technical arguments in favor as the primary metric, as 
discussed in draft-resnick-on-consensus, but validated the result using several 
other metrics (all listed in the slides) just to make sure that

A) no one was gaming the system, and

B) as chairs we weren’t missing some important consideration
For example, we counted # respondees not for declaring consensus per se, but to 
verify that there wasn’t degenerate cases such as are called out in 
draft-resnick-on-consensus.   And we were happy that we detected nothing of the 
sort.

7)      We went over our whole analysis with the AD before reporting results to 
the WG, to make sure we had support for our process and that our conclusions 
were valid.

8)      Finally, we then reported out to the WG

-Dave

From: Mark Townsley [mailto:[email protected]]
Sent: Wednesday, November 12, 2014 1:53 PM
To: [email protected] Group
Cc: Dave Thaler; Ted Lemon; Pierre PFISTER; Acee Lindem (acee)
Subject: Homenet Routing Protocol Discussion - 1500 Lehua Suite


We will continue our discussion on Routing Protocols in Homenet tomorrow at 
1500 in the Lehua Suite.  1500-1620 HST (Break + Thursday Afternoon Session II)

http://tools.ietf.org/agenda/91/venue/?room=lehua-suite

We expect this to continue to be an active discussion and debate. I've asked 
for a flipchart to be available for drawing diagrams, etc. Remote participation 
may be difficult, and for this we apologize. We will ask for note-takers and a 
jabber scribe (using [email protected]<mailto:[email protected]>) 
but as this is ad-hoc it will be on a best effort basis. This is where "conex" 
was going to meet before being cancelled, remote participants may be able to 
join an audio or video feed there.

Rough agenda:

At 1500, I'd like to ask Dave Thaler to give us his brain dump on what PCP did 
to help assist in their selection process. I know that Dave has to leave by 
1520 to participate in LWIG.

1520: Pierre Pfister will present an idea which was discussed among several 
members of the group in the hallway at the end of the meeting this morning

1605: Ray and I will direct the group towards a recommendation of next steps to 
be reported to the WG list.

1620: The rest of the IETF will be getting cookies and fruit, and we should be 
wrapping up.

1640: We lose the room, further discussions left to hallways and bars.

- Mark




_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet

Reply via email to