Skip to main content

Minutes for APPSAWG at IETF-93
minutes-93-appsawg-3

Meeting Minutes ART Area General Applications Working Group (appsawg) WG
Date and time 2015-07-20 07:00
Title Minutes for APPSAWG at IETF-93
State Active
Other versions plain text
Last updated 2015-08-14

minutes-93-appsawg-3
[07:06:08] <resnick> Recent RFCs published: 7505, 7565, 7578, 7595
[07:06:25] <resnick> rfc7001bis in the RFC Editor queue
[07:06:49] <resnick> markdown documents in IESG followup (i18n, etc, needs to
be resolved). [07:07:20] <resnick> 3 docs in development: -file-scheme,
-http-problem, -mdn-3798bis [07:07:36] <resnick> MNot speaks to -http-problem.
[07:08:00] <resnick> Got lots of feedback from W3C TAG. Still need to deal with
extensibility. [07:08:50] <resnick> Deciding on whether to add an extension
bucket to the JSON, or just bump the media type when you want to extend. Some
favor for the latter. [07:09:30] <resnick> Martin Thomson: Sounds fine, but
people are going to do extensions without bumping media type, so not a big
deal. [07:09:54] <resnick> (Bucket is like "x-", which is bad) [07:10:44]
<resnick> No other comments. Sounds like no need to bring it back to the WG.
New version forthcoming. [07:11:27] <resnick> Alexey on -mdn-3798bis: Still
working on grammar a bit. Still need to work through open issues. [07:11:58]
<resnick> Murray on -file-scheme: Sounds like one more round of feedback, then
WGLC. [07:12:34] <resnick> Barry on rechartering APPSAWG: [07:13:50] <resnick>
Looking to recharter APPSAWG to act more like DISPATCH; less handling of
documents. [07:14:31] <resnick> Finding homes for documents instead of handling
documents on its own (with a few easy exceptions). [07:15:21] <resnick> The ADs
are still figuring out whether to do a ART-wide area WG. [07:16:25] <resnick>
Dave Crocker: This was supposed to be a WG for documents that can't find homes.
What changed? [07:17:21] <resnick> Barry: We've gotten better about spinning up
new WGs quickly, so that's less of an issue. But there's also worries about
review here (both things not getting airtime, and getting buried in noise).
[07:17:40] <resnick> Alissa: Also possible for the decision for AD-shepherded
document. [07:17:52] <resnick> Murray: Can it go to ISE? [07:18:28] <resnick>
Alissa/Ben: They haven't, but DISPATCH can always say "No", which can end up
with things to ISE. [07:18:52] <resnick> Ben: DISPATCH also has gotten better
at spinning up quick focused WGs. [07:19:26] <resnick> Barry: And to repeat,
trying to figure out how to make area more cohesive. [07:19:29] <Keith Moore>
To be fair, the RAI stuff never was in APPS. [07:20:21] <resnick> Ben on the
new area: [07:20:48] <resnick> Talking about what RAI was. [07:21:05] <Barry
Leiba> Keith: Some was.  Geopriv, for example. [07:21:29] <resnick> Previously,
there was SIPPING. Moved to the DISPATCH process. [07:21:42] <rbarnes > "the
liver of RAI" [07:21:55] <resnick> (Slides stolen from 2013 tutorials. Will be
updating on the fly.) [07:22:37] <resnick> Not "real-time" in the computer
science sense. We mean "delay sensitive interpersonal communication".
[07:23:16] <resnick> Focused on building blocks. (See slides for details)
[07:23:46] <resnick> RTP, SIP, SDP, SIMPLE/XMPP, RTCWEB, GEOPRIV, ECRIT, CLUE,
PERC. [07:24:44] <resnick> RTP caries time-dependent data across jittery nets.
[07:25:01] <resnick> SIP is about Rendezvous and Negotiation. [07:25:25]
<resnick> Pretty picture of SIPing. [07:26:33] <resnick> SDP [07:26:40]
<resnick> RTCWEB [07:27:09] resnick gets bored re-typing things on slides. Go
look here: RAI WGs Overview
<https://www.ietf.org/proceedings/93/slides/slides-93-appsawg-4.pdf> [07:27:13]
<resnick> RAI WGs Overview
<https://www.ietf.org/proceedings/93/slides/slides-93-appsawg-4.pdf> [07:27:42]
<resnick> Telepresence. CLUE, "immersive" conferencing. [07:27:55] <resnick>
Pretty picture of Telepresence. [07:28:18] <resnick> Presence and Messaging
space. [07:29:29] Mary Barnes joins the room [07:30:03] <resnick> Calling party
identify (STIR) [07:30:23] <resnick> (Contrary to the slides, STIR is
well-underway) [07:31:13] <resnick> Emergency Services [07:31:30] <resnick>
(Works with other SDOs) [07:31:45] <resnick> DISPATCH - Assessing new work
[07:32:46] <resnick> Map of WGs not-very-pretty picture. [07:33:21] <resnick>
More map. [07:33:53] <resnick> (SIPCORE belongs on the previous slide)
[07:34:17] <resnick> More background slides; go look for yourself. [07:34:19]
<resnick> Q's?

Quick review of new Working Groups:

[07:35:27] <resnick> Ben: PERC, MODERN, NETVC are also new groups.
[07:35:59] <resnick> Ted Hardie: ACME - Trying to make it easy to
install/update certs. [07:36:09] <resnick> Sounds very SEC-sy. [07:36:10]
<Scott Hollenbeck> ...and LAGER, though not meeting this week [07:36:52]
<resnick> Matt Miller: JSONBIS - coordinating with ECMA and getting JSON to
Standard. Only meeting on the list. [07:37:19] <resnick> Alexey: IMAPAPND -
Extension to get big things uploaded to IMAP. Meeting on the list. [07:38:22]
<resnick> Murray - DBOUND: Figure out how to find relationships between domain
names. 17:40 today. [07:39:48] <resnick> Marc Blanchet: LAGER is about
beer^h^h^h^h a file format to describe a list of codepoints (i.e., text),
especially for DNS registries. [07:40:33] <resnick> Adam Roach: NETVC is doing
a royalty free video codec, similar to OPUS for audio. [07:40:54] <resnick>
Richard Barnes: PERC is about encrypted conferences that scale. [07:42:05]
<resnick> Henning Schulzrinne: MODERN is for managing phone numbers.

Quick review of BOFs:

[07:43:27] <resnick> Mark Nottingham: CAPPORT BOF - Trying to improve the
captive portal problem. [07:44:44] <resnick> Ted Hardie: Bar BOF on use of UDP
as substrate. Wed 20:20 Congress I [07:45:08] <resnick> Cullen Jennings: DETNET
BOF - Scheduling timeslots for packets. This PM [07:45:44] <resnick> Robert
Sparks: EDUNEXT BOF - Thinking about doing different things than tutorials,
different kinds of trainings. This PM. [07:46:41] <resnick> Leif Johanson -
SCIM bar bof (missed the time and place) [07:53:29] <Barry Leiba> SCIM
location, from "the list:": "The last session on Monday ends at 19:50, so let's
meet 8:15-ish at the Prague Beer Museum, Dlouhá 46, Prague 1" …
http://www.praguebeermuseum.com/en [07:48:08] <resnick> Linda Dunbar: I2NSF BOF
- Virtualized security functions use by multiple vendors. Tuesday 13:00.
Applications might want to be able to get at incident reports.

[07:49:01] <resnick> Alexey speaks to IMAP4rev2
[07:49:22] <resnick> Short History of IMAP.
[07:49:46] <resnick> IMAP4, revision 2
<https://www.ietf.org/proceedings/93/slides/slides-93-appsawg-10.pdf>
[07:50:15] <resnick> Goals - general. [07:51:11] <resnick> Goals - fixes
[07:51:48] <resnick> (I will presume if people want more details here because
they can't download the slides, they will say so.) [07:51:54] <resnick> Goals -
cleanup [07:53:03] <resnick> Goals - improvements [07:53:39] <resnick>
Challenges [07:55:47] <resnick> What I've done so far [07:56:01] <resnick>
What's next? [07:57:43] <resnick> Pete Resnick: Why not charter now? [07:58:12]
<resnick> Barry: Because it might distract IMAPAPND. But OK with me. [07:58:40]
<resnick> Dave Crocker: Should get a dedicated mailing list. [07:59:05]
<resnick> Barry: But we need the people on imapext [07:59:23] <Keith Moore> IMO
there is such a thing as too many mailing lists, too much fragmentation of
discussions. [07:59:30] <resnick> Dave: But this is a focused effort. Make it a
separte list. [07:59:42] <resnick> Barry: Suggest you draft a charter and
create a list. [07:59:52] <resnick> Done.

[08:00:37] <resnick> Cullen Jennings on WebRTC Enterprise Firewall Traversal
[08:00:56] <resnick> Skipping to 4th slide (P2P Media)
[08:01:14] <resnick> Firewalls and WebRTC Traffic
<https://www.ietf.org/proceedings/93/slides/slides-93-appsawg-11.pdf>
[08:02:19] <resnick> Back 2 slides to "Proposed Firewall Algorithm for WebRTC"
[08:03:33] <resnick> Looking for input on algorithms, but also trying to figure
out who the experts are here. [08:04:12] <resnick> Eric Rescorla: There are
still some problems with the algorithm [08:04:24] <Keith Moore> Did ekr just
say "firewall" and "totally reasonable" in the same sentence? [08:05:08] <Barry
Leiba> I think the angle was more "totally reasonable approach to dealing with
firewalls", which isn't quiiiiiite the same thing. [08:05:12] <resnick> Linda
Dunbar: Sounds like I2NSF doing similar things. Is it? [08:05:46] <resnick>
Cullen: This is more old-school, the firewall is working independently.
[08:06:24] <resnick> Linda: From the firewall perspective, it's just a signal
coming in. So, this is somewhat similar. Are there particular semantics?
[08:06:35] <resnick> Cullen: Yeah, that's what we want to have a technical
discussion about. [08:07:11] <resnick> Cullen: This is a strawman proposal to
address that. Discussing this on the BEHAVE list. [08:07:34] <resnick> Linda:
Having this discussion on I2NS too. We should coordinate. [08:07:40] <resnick>
Cullen: Sounds good. [08:08:04] <resnick> Eric Rescorla: Key seems to be do
this at the firewall without having to change the clients. Right? [08:08:14]
<resnick> Cullen: Right for WebRTC, not for SIP. [08:08:36] <resnick> Cullen:
Even SIP clients that do ICE don't do everything needed for this. [08:09:30]
<resnick> David Black: STUN and friends are pin-hole punchers. WebRTC won't
work without that. Since we're doing that, we want this added thing. Caution.
[08:09:45] <resnick> Last slide: What to do about media hiding in HTTPS
[08:12:52] <resnick> Adam Roach: One additional consideration - Proxies are for
privacy. Need to be cognizant of what that means for running WebRTC over the
proxy. [08:13:36] <resnick> Ted Hardie: When you say request...? [08:13:47]
<resnick> Cullen: They mean "outbound packet" [08:13:57] <resnick> Ted: That
ain't going to fly in HTTP2. [08:14:18] <resnick> Ted: You need a dedicated
mailing list and some Prozac. [08:14:40] <resnick> Martin Thomson: Doing this
because you are unhappy with the current solution? [08:15:00] <resnick> Cullen:
I've gotten feedback from vendors. [08:15:18] <Eliot Lear> oh this is hogwash. 
the middlebox vendors have been largely ignored in that wg [08:15:32] <resnick>
Martin: Middlebox vendors have participated [08:16:30] <Barry Leiba> Eliot, to
you want to be channeled at the mic? [08:16:31] <resnick> Christian Huitema:
Transparency vs. Control. Going on the control side is going to screw with
transparency and privacy. Going to make firewalls worse than they are.
[08:16:38] <Keith Moore> +1 to Christian Huitema's comment [08:17:00] <Eliot
Lear> sure- but i am not responding to the content of the draft, just martin's
statement [08:17:31] <Barry Leiba> in q [08:18:04] <resnick> Eric Rescorla:
This is phrased less helpful than it might. Surprised people think it's not
possible to block voice traffic; of course firewalls can do that. [08:18:44]
<resnick> There are two concern that admins have: 1. I don't want voice/videoc
calls at all. 2. I wish they'd choose a different proxy. [08:19:06] <resnick>
Easier way to say, "Here's a better proxy" is what we want. [08:20:16]
<resnick> Or a signal to say, "I don't want voice traffic" [08:20:27] <resnick>
Should be some signalling to do that. [08:22:33] <resnick> Dave Crocker: Heard,
"Try to get firewalls to do things the same way", but also "Let's try to make
the hack work better". Seems like a cleaner solution is better than the fix to
the hack. [08:23:05] <Keith Moore> The goal in all cases should be to make the
network's behavior (including that of firewalls) predictable. [08:23:41]
<resnick> Cullen: Two parts: What can you do in enterprise and what can you do
in residential. Residential we're stuck. Enterprise you could change, but
getting people to change the settings is hard. [08:25:40] <resnick> Joe
Hildebrand: Dave sounded like we needed an architectural framework. One day we
might have something like that (e.g., potato-ish things), but as you learn
about this, bring that back to stack-evolution program in the IAB. [08:26:26]
<resnick> Aaron Falk: This looks like a new killer app is stressing the
infrastructure, so either fix or block the app. But that's the proxy's problem,
not the app. [08:27:26] <resnick> Eric Rescorla: HTTP CONNECT is a terrible
mechanism to push voice traffic through the firewall. It would be really nice
for there to be a mechanism for apps to do so. [08:27:27] <Keith Moore> I keep
wishing that apps had a reliable way of pointing out exactly what about the
network is keeping the app from working.   e.g. this app is being prevented
from working by a misconfigured firewall, and the administrator's address is
joe@example.com. [08:28:37] <hildjj> for those that didn't get it: "potato-ish"
is a reference to draft-hildebrand-spud-prototype.  This is NOT a protocol you
can use here, it's a prototype that's an input into the IAB stack evolution
program.  One of the things we're exploring there, Keith, is explicit
mechanisms for path-to-app communication (such as errors) in a
protocol-independent way. [08:28:40] <resnick> Cullen: I'll send email to HTTP
list; interested to get an answer to EKR's issues.

[08:30:16] <resnick> WG Reports
[08:31:38] <resnick> Dave Crocker: ARCMEDIA - Developing a top-level media type
for archives. Chartered in February. Almost no activity. Probably appropriate
to shut it down. Comments? [08:31:58] <resnick> Crickets.... [08:32:35]
<resnick> Mark Nottingham: HTTPBIS - HTTP2 shipped, but there are more things
in the pipeline. [08:33:06] <resnick> Cullen Jennings: RTCWEB is getting close
to done on the key documents. Pretty boring stage, but need reviewers.
[08:33:48] <resnick> Mo Zanaty: NETVC - Looking to get people from assorted
affiliations to get involved. [08:34:19] <resnick> Alex Mayhoffer: DRINKS is
down to just me. Finishing up. [08:34:52] <resnick> Alissa: There's a bunch of
groups on the list that are closing soon. At least 6.

[08:35:53] <resnick> Mark Nottingham and Wendy Seltzer on W3C update.
[08:36:28] <resnick> W3C/IETF relationship quiet for the moment. Nothing in
RTCWeb/WebRTC problematic at the moment. [08:36:59] <resnick> Liaising is
tending toward pointing W3C people to assorted IETF stuff. [08:37:45] <resnick>
(Wendy lists many technologies) [08:37:53] <resnick> Cullen Jennings: Web of
Things? [08:38:20] <resnick> Wendy: There's a Web of Things Interest Group.
[08:38:39] <resnick> Mark: Webappsec seems to be the locus. [08:39:03]
<resnick> TPAC is in Sapporo the week before Yokohama. [08:39:46] <resnick>
Working on the details of how to coordinate. [08:40:38] <resnick> TAG met last
week in Berlin. Talking about e2e encryption (without back doors) and [thing
Pete missed]. [08:40:39] <hildjj> http://www.w3.org/2015/10/TPAC/ [08:41:25]
<resnick> Wendy: HTML activities being restructured. New Web Incubator
Community Group <https://www.w3.org/community/wicg/> [08:48:55] <wseltzer>
http://wicg.io/

[08:43:00] <resnick> Takehito Akagiri on Mail Divide
[08:43:10] <resnick> Mail Divide Framework
<https://www.ietf.org/proceedings/93/slides/slides-93-appsawg-7.pdf> [08:45:05]
resnick paying attention to slides, but he's going through them slide for
slide, so nothing to add here.

[08:50:16] <resnick> Personal Cloud Storage Sync Protocol
<https://www.ietf.org/proceedings/93/slides/slides-93-appsawg-8.pdf> [08:52:30]
<resnick> (Yong Cui) [09:01:15] <resnick> Joe Hildebrand: Any of the vendors
interested? [09:01:22] <resnick> Yong: No. [09:01:45] <resnick> Eric Rescorla:
There are existing protocols, but none of the vendors used them. [09:01:55]
<John Klensin> Mic: Addition to Joe's question: there are a bunch of vendors
who are selling "personal" network storage devices designed for the same sort
of purposes, but under enterprise or indivudal control, rather than third-party
vendors -- anyone talking with them?  Seems that we shouldn't need two sets of
protocols. [09:02:03] <Barry Leiba> in q [09:02:15] <resnick> Mark Nottingham:
Doesn't sound like vendors are interested. Maybe just a sync protocol is worth
it. Need to be realistic. [09:03:14] <resnick> Leif Johanson: Supporting Mark.
It's not the sync that locks in. The vendors won't listen to us. Need to go
talk to them directly. [09:03:49] <resnick> John Klensin channeled. [09:04:41]
<resnick> Mark Nottingham: Bad that people's data put into proprietary silos.
But need to take IM lessons to heart. [09:06:28] <hildjj> market incentives
drive existing vendors away from standards in order to get brand awareness on
your desktop.  think about other incentives you could offer them [09:06:57]
<rbarnes > contrary to some peoples' beliefs, standards bodies can't force
anyone to do anything [09:07:24] <resnick> Cullen Jennings: This needs to be
BOFed. [09:08:21] <hildjj> It needs to be BoF'd, but it would be surprising to
me if he could get a successful one put together by Yokohama.  I'm prepared to
be impressed and surprised however. [09:08:45] <resnick> Bob Briscoe on Data
Centre to the Home - Ultra Low Latency for All. Come to AQM or Bits n' Bytes..
[09:10:26] <John Klensin> @hildjj: That is one of the reasons I'm interested in
the individual/enterprise devices.  Their vendors  might have incentives to
play and, if they played together, it might help stimulate the big actors. for
all oft the second source reasons.