[04:02:26] --- hartmans@jis.mit.edu/owl has joined
[04:05:19] --- patchvonbraun has joined
[04:05:33] --- Stephen Farrell has joined
[04:05:46] --- Stephen Farrell has left
[04:06:10] --- menno pieters has joined
[04:06:15] --- Fred has joined
[04:06:44] --- clancy has joined
[04:07:17] --- Stephen Farrell has joined
[04:08:02] --- tlyu has joined
[04:08:15] --- raeburn@mit.edu has joined
[04:08:26] --- kivinen has joined
[04:08:40] --- vidya has joined
[04:08:54] --- jhutz@jis.mit.edu/owl has joined
[04:09:07] <jhutz@jis.mit.edu/owl> Someone want to fix the topic?
[04:09:17] --- weiler has joined
[04:09:52] <jhutz@jis.mit.edu/owl> We are on channel 5: http://videolab.uoregon.edu/events/ietf/ietf685.m3u
[04:10:11] --- dumdidum has joined
[04:11:12] --- saag has joined
[04:11:34] --- rlbob has joined
[04:13:55] --- ldondeti has joined
[04:15:01] <jhutz@jis.mit.edu/owl> OK; who gave their nick as "saag" ?
[04:16:16] --- tlr has joined
[04:17:54] --- rjaksa has joined
[04:24:32] --- ekr has joined
[04:25:41] <ekr> Just as a clarification of the SIP issue, the concept here is that (as with a mail server) the SIP proxy gets to decide who gets to have a given address under its name. So, you can sign an assertion that says "this message came from user X" before forwarding it to a different proxy
[04:25:47] --- fenton@jabber.org has joined
[04:25:52] --- fenton@jabber.org has left
[04:26:04] --- fenton has joined
[04:28:25] --- stjepan.gros has joined
[04:31:12] --- dumdidum has left: Replaced by new connection
[04:31:52] --- stjepan.gros has left
[04:33:17] --- uli.m has joined
[04:39:24] <rlbob> ekr: so the SIP idea is that RPs will get messages signed by SIP proxies saying "here's a message from user X", and have to decide whether the asserting entity (a) is a SIP proxy and (b) can make claims on behalf of that user?
[04:40:26] --- roy has joined
[04:41:30] <ekr> Well, pretty much by definitition it's a SIP proxy since it's sending you SIP :). In terms of deciding whether it is allowed to be a proxy, you examine the cert, and the question of whether it can make claims is that by definition SIP proxies are allowed to do so
[04:41:30] --- weiler has left
[04:42:49] <fenton> Can any proxy assert any identity, or does there need to be some relationship between the identity and the proxy?
[04:42:53] <rlbob> hmm, I thought one of the design questions was having some indication in the cert that this thing is a SIP proxy, as opposed to any old thing that's putting out signed SIP messages
[04:43:00] <tlyu> i hope i wasn't too unintelligible... not enough sleep
[04:43:55] <rlbob> tlyu: just unintelligible enough :)
[04:44:08] --- nico has joined
[04:44:44] <ekr> The domain name of the proxy (i.e., in the cert) must match the RHS of the SIP URI.
[04:45:35] <fenton> sounds vaguely DKIM-like :-)
[04:45:37] <ekr> rlbob: yeah, that is the idea.
[04:46:13] <ekr> Fenton: sort of, but you're also attesting the LHS, which, as I understand it, DKIM explicitly disclaims
[04:49:35] <rlbob> I ask because in SAML, where "identity providers" do things very similar to these asserting SIP proxies, there is so-called "metadata" describing these providers in potentially great detail, including cert names and/or keys; kind of inside-out from what's being proposed for SIP
[04:49:52] <fenton> ekr: there aren't explicit rules in DKIM about what the MTA can sign, but generally it shouldn't sign (take responsibility for) a message unless it believes the LHS. And the i= tag is an indication of specifically which address (including LHS) it's talking about.
[04:51:13] --- roy has left
[04:51:33] <ekr> Well, there are only so many ways to do things, obviously, so yeah, it is a little like DKIM, but it's explicitly for authentication
[04:51:57] <fenton> cool.
[04:53:45] --- weiler has joined
[04:54:26] <tlr> apologies for not being there in person
[04:54:34] --- stefan.santesson has joined
[04:54:54] --- HannesTschofenig has joined
[04:55:02] <jhutz@jis.mit.edu/owl> Maybe we should have arranged to put you on a second screen, and let you present directly.
[04:55:13] --- cabo has joined
[04:55:38] --- stjepan.gros has joined
[04:57:21] <jhutz@jis.mit.edu/owl> Are you listening to the audio?
[04:57:34] --- uli.m has left
[04:57:36] <tlr> It is overview document, starting on the recommendations soon.
[04:57:40] <tlr> Sonner than June, actually.
[04:58:30] <tlr> Yes. And CMU is actually on the group.
[04:59:42] <tlr> Scope is explicitly "no new protocol work here"
[05:00:06] --- uli.m has joined
[05:01:57] --- mtcarrasco has joined
[05:02:13] <tlr> Yes, precisely. That's in scope for the Forms WG. Just got chartered.
[05:02:39] <tlr> Lots of HTML forms work will be part of a joint TF between HTML and Forms WGs.
[05:02:41] --- kuro has joined
[05:03:21] <hartmans@jis.mit.edu/owl> Hmm. I should get Leif to look at the forms stuff and make sure his auth stuff could work well within that
[05:03:46] <rlbob> which auth stuff?
[05:05:00] <hartmans@jis.mit.edu/owl> The stuff we sat around a table discussing at the last meeting
[05:05:01] <nico> rlbob: TLS chanell bindings for https with proxy traversal
[05:07:00] <tlr> Thanks Stephen!
[05:07:10] <rlbob> i don't see how that relates to Forms, unless you're thinking that some Forms-based thing would be made into a real HTTP authentication method
[05:07:17] <nico> but more importantly, here, I think Sam is referring to the HTTP GSS + TLS CB thing
[05:07:50] <nico> see draft-johansson-tls-*
[05:08:02] <tlr> We have at least 8 weeks' notice for workshops, typically more, so having the workshop before July is not realistic.
[05:08:11] <nico> rlbob: that's what the slides we just saw seemed to imply!
[05:08:45] <tlr> they are open
[05:08:48] <hartmans@jis.mit.edu/owl> For gss auth to be useful you need to be able to trigger it from a form.
[05:08:52] <rlbob> a subtle implication, I guess :)
[05:09:05] <nico> actually, it was explicit
[05:09:13] --- Fred has left
[05:09:28] <nico> form extensions to specify protection-like attributes
[05:09:38] <nico> it didn't say authentication on the slide
[05:09:53] <nico> but it seems to follow, no?
[05:10:21] --- Stephen Farrell has left: Replaced by new connection
[05:10:22] --- Stephen Farrell has joined
[05:11:12] <nico> hmmm, actually, what we want is a way to say "do this for authentication" where "this" can be a form submission with username/password over TLS, or where it could be some HTTP auth mech, possibly using username/password from a form, possibly not
[05:11:14] <hartmans@jis.mit.edu/owl> No, I would not assume it follows
[05:11:33] <nico> we need to participate and comment then
[05:11:37] <hartmans@jis.mit.edu/owl> But I don't have time to get involved in a W3C WG even if I could.
[05:12:00] <nico> not even to fire and forget? :)
[05:12:52] <hartmans@jis.mit.edu/owl> W3C does not support fire and forget
[05:12:58] <hartmans@jis.mit.edu/owl> We don't either
[05:14:17] <tlr> One useful way to contribute would be to do a short write-up of the requirements that you guys have in mind.
[05:14:28] <jhutz@jis.mit.edu/owl> Sure we do. We just don't promise reliability.
[05:15:39] <hartmans@jis.mit.edu/owl> For at least part of this I've done that and am requesting Lisa do a last call. If that goes forward we'll send it your way
[05:16:09] <tlr> Great, thanks!
[05:19:07] <nico> sam: thanks
[05:20:12] --- raeburn@mit.edu has left
[05:20:42] <rlbob> sam: you mean draft-hartman-webauth-phishing-03.txt right?
[05:22:18] <jhutz@jis.mit.edu/owl> Sure, the IETF would be interested in working on such a profile. We love a good argument.
[05:23:05] --- fenton has left: Replaced by new connection
[05:25:19] --- fparent@jabber.org has joined
[05:26:15] <hartmans@jis.mit.edu/owl> rlbob: Yes.
[05:26:32] <hartmans@jis.mit.edu/owl> I think I've finally dealt with ekr and Ben's comments.
[05:28:44] --- HannesTschofenig has left: Replaced by new connection
[05:33:38] <hartmans@jis.mit.edu/owl> Russ's slides should now be on the website
[05:33:52] <hartmans@jis.mit.edu/owl> Well, if the web server ever responds
[05:39:17] <jhutz@jis.mit.edu/owl> I was able to upload something a moment ago.
[05:40:01] --- HannesTschofenig has joined
[05:41:33] <ekr> Oh, that's just too easy.
[05:41:45] <ekr> "Is this total nonsense or am I just confused? Can you clarify?"
[05:42:06] <Stephen Farrell> I don't think ekr is confused
[05:42:06] <hartmans@jis.mit.edu/owl> *laugh*
[05:43:22] <ekr> Actually, I haven't decided about this preso yet. I just thought it was easy to circumvent the restrictions :)
[05:43:59] --- geoff has joined
[05:45:47] <ekr> He's right that there are a lot of issues we didn't cover in 3552, but I'm not sure that this is the list I would choose.
[05:47:01] --- JeffH has joined
[05:47:56] <ekr> the reason I ask is that the problem was an API design problem.
[05:48:01] <ekr> They didn't get through the tamper
[05:48:35] <ekr> Which isn't to say it wasn't serious, but think of it as a software bug
[05:49:00] <Stephen Farrell> well, they did have too many interfaces defined - much easier to get tripped up then
[05:49:33] <ekr> Yeah.
[05:50:17] <ekr> It's actually an interesting question how hard it is to mount a head-on attack on a 4758
[05:50:28] <ekr> My understanding is that it
[05:50:32] <ekr> 's doable, but not easy.
[05:51:25] <ekr> I approve of women with guns
[05:51:27] <hartmans@jis.mit.edu/owl> Yeah. I think that he needs to describe what the scope of his threat model is. If he's saying that intarea needs a threat model and that this is close to the threat model they need, that may be true.
[05:52:00] <jhutz@jis.mit.edu/owl> So, looking at the IETF68 proceedings page a few minutes ago, SEC had both the largest number and largest percentage of "green" sessions (those for which agendas, presentations, and minutes had all been submitted).
[05:52:16] <hartmans@jis.mit.edu/owl> But I don't want to mess with this threat model for say webdav
[05:52:38] <ekr> Lakshminath is PC.
[05:52:51] <nico> lol
[05:52:54] <ekr> Is that a clarifying question?
[05:52:57] <ekr> Sam?
[05:53:19] <ekr> Hmm... That sounded like a comment
[05:54:35] <jhutz@jis.mit.edu/owl> It was a question. He made a comment to clarify the question. Besides, he's chairing the session, and so gets some leeway.
[05:54:41] --- JeffH has left: Logged out
[05:54:45] <ekr> heh.
[05:55:07] <hartmans@jis.mit.edu/owl> No, I screwed up. I thought he was done.
[05:55:08] <jhutz@jis.mit.edu/owl> When you are SEC AD... Well, we'd still prefer you were nice and let people finish their presentations before you started ripping them to shreds. :-)
[05:55:11] <ekr> One of my favorite clarifying questions is "where are you going with this?"
[05:55:28] <hartmans@jis.mit.edu/owl> This is too long.
[05:55:36] <nico> ekr: heh
[05:55:36] * tlr chuckles. That's "where do you want to go today", ekr
[05:56:13] <nico> unstructured indeed
[05:56:29] <ekr> I'm trying to figure out what the issue is here....
[05:56:30] --- raeburn@mit.edu has joined
[05:56:49] <weiler> hartman: you're the one with the hook....
[05:56:57] <jhutz@jis.mit.edu/owl> He's on the last "real" slide
[05:57:00] <nico> ah, proposed work
[05:57:10] <nico> but it is nebulous
[05:57:29] --- raeburn@mit.edu has left
[05:57:34] <jhutz@jis.mit.edu/owl> saag is no stranger to nebulous proposals
[05:57:38] --- raeburn@jis.mit.edu has joined
[05:57:41] --- slm has joined
[05:58:09] <ekr> I'm thinking I'm perhaps going to design my own bad SAAG talk.
[05:58:28] <ekr> Like the ideal one.
[05:58:32] <nico> but you cannot take half an hour giving it
[05:58:34] <ekr> But I gotta think what it would have.
[05:58:39] <nico> not at ekr speed
[05:58:42] --- dbh2 has joined
[05:58:43] <weiler> ekr: make sure you use pretty pictures
[05:58:53] <weiler> and a custom slide background. :-)
[05:59:22] <rlbob> i was impressed, i think Steven Farrell made it to perhaps 0.8 EKRs during his pitch
[05:59:24] <ekr> It would definitely contain this diagram
[05:59:26] <ekr> http://upload.wikimedia.org/wikipedia/commons/6/60/Ims_overview.png
[05:59:34] <nico> wwg -- women with guns, and witm -- woman in the middle. at least we got new acronyms!
[06:00:22] <ekr> I invite you to explain the difference between the P-CSCF, the I-CSCF, and the S-CSCF
[06:00:31] <Stephen Farrell> rlbob: so that's an EK then?
[06:02:11] <rlbob> um, I guess
[06:03:40] <rlbob> i think "transgendered person in the middle" may be a more appropriate threat ...
[06:04:21] <nico> to be really PC it might have to be PITM or HITM
[06:04:43] --- ldondeti has left: Disconnected.
[06:04:53] <nico> is Mallory always a female name in English?
[06:05:05] <nico> if not we could say MITM is Mallory in the middle
[06:05:06] <rlbob> could be either
[06:05:27] <jhutz@jis.mit.edu/owl> Yeah, could be either. Americans will probably assume it is female.
[06:05:54] <jhutz@jis.mit.edu/owl> ekr, this slide?
[06:06:40] <nico> ekr is at the mic
[06:08:31] <nico> we are have 23 minutes
[06:09:08] <jhutz@jis.mit.edu/owl> Hm; I wonder if I can have faster network if I disconnect the ethernet cable from this access point and plug it into my laptop instead. :-)
[06:10:12] --- raeburn@jis.mit.edu has left: Disconnected
[06:10:48] <jhutz@jis.mit.edu/owl> Can we move to open mic?
[06:14:00] --- HannesTschofenig has left: Replaced by new connection
[06:14:51] <ekr> So, to make the situation even worse, there are systems with nondeterministic security models.
[06:14:56] <ekr> cf. security for DHTs.
[06:15:19] <rlbob> dhts?
[06:15:43] <nico> distributed hash tables?
[06:15:55] <rlbob> ah
[06:16:10] <nico> I'm asking
[06:16:39] <nico> but yeah, I think distributed hash tables may have non-deterministic security models
[06:17:09] <rlbob> to go along with non-deterministic behavior in general ...
[06:17:27] <ekr> yeah, distributed hash tables.
[06:17:38] <nico> rlbob: sure
[06:18:56] <ekr> So, building systems which degrade gracefully is really hard
[06:21:27] <ekr> Dan Harkins: "Stop The Madness!!!!"
[06:21:46] --- uli.m has left
[06:22:03] --- menno pieters has left
[06:22:50] --- mtcarrasco has left
[06:23:25] --- stjepan.gros has left
[06:26:02] --- dbh2 has left
[06:26:09] --- rlbob has left
[06:26:17] <weiler> No one has a stick?
[06:26:28] <weiler> there will be playing of "fetch"?
[06:26:50] --- rjaksa has left
[06:28:03] <weiler> BTNS!
[06:30:49] <nico> samw: indeed!
[06:30:50] --- clancy has left: Logged out
[06:30:58] --- vidya has left
[06:31:02] --- fparent@jabber.org has left
[06:31:50] --- kuro has left
[06:32:17] --- weiler has left
[06:32:56] --- Stephen Farrell has left
[06:33:00] --- nico has left
[06:33:18] --- kivinen has left
[06:33:27] --- patchvonbraun has left
[06:33:31] --- tlr has left
[06:33:41] --- slm has left
[06:33:51] --- saag has left
[06:41:40] --- jhutz@jis.mit.edu/owl has left: Disconnected
[06:51:52] --- tlyu has left
[07:14:26] --- geoff has left
[07:19:07] --- cabo has left
[07:28:30] --- ekr has left
[08:17:38] --- mrex has joined
[08:25:16] --- mrex has left
[11:04:39] --- stefan.santesson has left
[14:32:24] --- stefan.santesson has joined