[dispatch] IETF 99 DISPATCH meeting notes - summary

["A. Jean Mahoney" <[e-mail]>]

Hi all,

Here are my notes all summarized and compact.

Thanks!

Jean

---------------------------------------------------------------------

DISPATCH WG Session

IETF 99 — Prague, Czech Republic
July 17, 2017  09:30-11:30, Congress III


---------------------------------------------------------------------
09:30 Administrivia - Chairs (5 min)

Mary Barnes presented chair slides.
Note takers: John Levine, Jean Mahoney.
Jabber scribe: Rich Salz

Mary reminded people that charter proposals don't have to be formal. Ben 
Campbell announced that CELLAR was actually meeting this week. Murray 
announced that APPSAWG is closed.


---------------------------------------------------------------------
09:40 Web Packaging - Jeffrey Yasskin (30 min)
 
https://www.ietf.org/proceedings/99/slides/slides-99-dispatch-web-packaging-00.pdf
       https://github.com/WICG/webpackage

Mark Nottingham felt that two aspects of the problem could be worked on 
- A format to persist HTTP requests/responses, and the ability to assert 
that a request/response pair is from an authority (signatures). Phillip 
Hallam-Baker wanted a proper specification for MHTML first. Magnus 
Westerlund suggested that Jeffrey read about out-of-band encoding. 
Martin Thomson declared that the problem had been pitched numerous 
times. Eric Rescorla pointed out that there were security issues with 
allowing an HTTPS signature to span weeks. Daniel Gillmor brought up the 
flavors of Linux distributions and that they still don't handle 
dependencies well and they were centralized, unlike this solution. 
Patrick McManus did not feel the IETF was the right venue for this. Mark 
would love to discuss it further and felt it could go to HTTPBIS working 
group, just not now. It needed a careful description of use cases and 
solutions to those use cases. The W3C should also be consulted before 
the IETF took the work on. Daniel said that there needed to be a better 
understanding of what needed to be signed. Ben Schwartz supported the 
work. Paul Hoffman said that people misinterpret the meaning of signed 
packages, and wanted to know if people could use them. Eric wanted to 
start with just the use cases and the requirements. People in the Jabber 
room said that it should have a BOF.

Cullen asked how many people were interested in working on this problem. 
About 10 hands raised.

Murray said the discussion could continue on the art mailing list. 
Cullen said if the discussion got too heavy, a new mailing list could be 
spun up.

ACTION: Continue discussion on ART mailing list.


---------------------------------------------------------------------
10:10 DNS Over HTTPS - Paul Hoffman (40 min)
 
https://www.ietf.org/proceedings/99/slides/slides-99-dispatch-dns-over-https-00.pdf
       draft-hoffman-dispatch-dns-over-https


Magnus wanted to know how the resolver was controlled, and who has 
control of which HTTP server responds, and had concerns about and 
privacy. Patrick said that discovery problems are out of scope, but it 
provides an improvement for privacy. Mark thought the draft was good, 
well-scoped and agreed on punting on the use right now. He said that the 
draft should take care not to call http1 legacy. It is still in use. 
Mark also raised the issue of the peers downgrading to http1 - what 
would happen? Dave Lawrence like the draft so far but did not agree with 
dropping the use cases. (Dave also mentioned that there was a competing 
idea [which the note taker failed capture] that should be acknowledged 
in the draft. Ted Hardie wanted to know why DPRIVE wouldn't work. Paul 
said that this solution wasn't simply about privacy and taking it to 
DPRIVE would require rechartering. Keith Moore felt that this solution 
forked DNS, that there would be web-specific interfaces. John Klensin 
asked about content negotiation. Paul clarified that it was for format 
(like wire format or JSON), not different names. Yoav Nir had a question 
[missed by note taker], that Paul responded that DPRIVE traffic could be 
identified and blocked. With this solution, DNS traffic could not be 
identified. Nir also brought up the fact that intermediaries could 
downgrade http2 connections to http1, and that the draft should discuss 
what to do there. Nir asked when was POST used. Patrick responded that 
POST enables media negotiation in a better way, while GET had better 
caching properties, but it was up for discussion. Daniel supported the 
work, and recommended having something like doe [?]. Daniel pointed out 
that you can't validate DNSSEC over JSON, only over wire format. Paul 
said that wire format would be mandatory to implement. Daniel was also 
glad that discovery was not in scope, that http2 was specified, but 
wasn't sure about arbitrary endpoints and mentioned something about 
.wellknown. Eric was sad that this was being discussed just as DPRIVE 
was finishing, and wanted to know the settings in which HTTP and DNS 
could be mixed. He didn't feel that this was a good way of hiding 
traffic. Eric wanted to see use cases. Bron Gondwana pointed out that 
Thunderbird doesn't do DNS lookups and that wire format would be more 
difficult to implement, unlike JSON [he had another point that the note 
taker missed.]  Ben Schwartz pointed out there were two independent 
attempts to solve this issue and it would be good to standardize. 
Jonathan asked a question about ports and wanted to know if a java 
script implementation could do this without the browser knowing, the 
answer is yes.

Cullen asked for a show of hands of who was willing to read drafts and 
do work. Many, many hands raised.

ACTIONS: Art ADs will talk with Int and Sec ADs about DPRIVE, and will 
look for a place for it.

---------------------------------------------------------------------
---------------------------------------------------------------------

ARTAREA Session

---------------------------------------------------------------------
10:50 BoF Summaries - various artists (5 min)

BANANA - BANdwidth Aggregation for interNet Access (int)
IDEAS - IDentity Enabled Networks (rtg)
IASA20 - IASA 2.0 (gen)
NETSLICING - Network slicing (ops)

No volunteers to provide summaries.


---------------------------------------------------------------------
10:55 New Working Group Summaries -various artists (5 min)

DCRUP - DKIM crypto update

The WG is sharing time with DMARC and is meeting at 10:30.


---------------------------------------------------------------------
11:00 Using URIs With Multiple Transport Stacks - Dave Thaler (15 min)
       draft-thaler-appsawg-multi-transport-uris


Dave highlighted that the point of the draft was not to make 
recommendations, just capture the ways the problem has been solved and 
detail the trade offs. Phillip wanted a draft that provided a single 
solution. Keith Moore pointed out there was issues with changing the 
meaning of identifier schemes, and that layers of indirection are added 
later. Mark felt that it was appropriate to give this kind of advice and 
said that a larger problem was how to identify protocols. Barry Leiba 
felt that maybe it should be a BCP not just an informational doc.

ACTION: Get input from the community on the art mailing list.


---------------------------------------------------------------------
11:15 Hybrid Video Content - Roni Even (10 min)
       draft-huang-dispatch-hybrid-video-delivery
 
https://www.ietf.org/proceedings/99/slides/slides-99-dispatch-video-delivery-in-hybrid-network-00.pdf

Magnus felt the draft was too fluffy, covering several different 
problems, and it overlapped with the work in other fora. Keith thought 
that there may be a role for collaboration between MPEG and IETF.

Cullen asked, of the people who build these systems, who would be 
willing to implement a new IETF solution and who would participate. 3 
hands raised.

ACTION: Chairs to encourage discussion the mailing list.


---------------------------------------------------------------------
11:25 Open Microphone/AOB

Volker Birk from pretty Easy privacy (pEp) foundation spoke briefly 
about his foundation's opportunistic encryption application for chat and 
email. They are interested in making it an open standard. More 
information can be found here: 
https://www.ietf.org/mail-archive/web/saag/current/msg07789.html

Murray recommended that they submit drafts and post to the art list. 
Matthew Miller suggested that they also talk to the XMPP standards 
foundation.