Tuesday, 15 November 2011< ^ >
Adam Roach has set the subject to: Obsfucation Suckitude, IETF 81, Quebec
Room Configuration

[06:51:57] Martin Thomson joins the room
[06:52:32] <Martin Thomson> /help
[07:23:18] Jonathan Lennox joins the room
[07:23:19] tsuichi joins the room
[07:25:52] genki_yasutaka joins the room
[07:26:23] <Martin Thomson> yes
[07:26:29] ray joins the room
[07:26:36] <Martin Thomson> I am in the jabber room
[07:26:39] <ray> Hey!
[07:26:49] <Martin Thomson> hey!
[07:27:32] RjS joins the room
[07:27:51] richard.barnes joins the room
[07:28:10] <richard.barnes> hey all, i'll be your co-chair today
[07:29:55] <ray> who here is remote?
[07:30:56] <Martin Thomson> I'm at home
[07:31:48] Randall Gellens joins the room
[07:34:01] suzukisn joins the room
[07:35:23] Kaveh Ranjbar joins the room
[07:36:59] <Martin Thomson> mic: even if you have an algorithm that incorporates multiple values over time, the attacker can always beat you
[07:37:27] <Martin Thomson> I hope that's clear from the writeup
[07:37:46] <Jonathan Lennox> Who's doing mic relay?
[07:37:50] <Jonathan Lennox> I can if no one else is...
[07:38:47] <ray> I kinda was, but I wasn't paying enough attention... sorry
[07:39:12] <Jonathan Lennox> If you can do it going forward, I won't have to trip over Robert again. :-)
[07:39:15] <Martin Thomson> Thanks for the relay, and the quote for today :)
[07:40:16] <Martin Thomson> I haven't seen an implementation of what is in the draft
[07:42:14] <ray> @jl we just need to get Robert to relay - he's closest to the mic ;-)
[07:43:03] <Martin Thomson> For the record (and not the mic): on the implementation of the algorithm in geopriv-policy, there are a few underspecified parts to that algorithm that made it difficult for me to implement it
[07:43:25] <Martin Thomson> I wish Joe luck when he comes to animate it
[07:44:58] <Martin Thomson> I agree with the changes to the IANA registry, sounds good
[07:45:10] <Jonathan Lennox> Martin: was that for the mic?
[07:45:25] <Martin Thomson> Brian should probably know...and he isn't in here
[07:45:36] <Martin Thomson> so yes
[07:49:08] Randall Gellens leaves the room
[07:49:28] Randall Gellens joins the room
[07:50:29] <Martin Thomson> ship it!
[07:51:02] <Martin Thomson> use case: you want to use local reference systems for mapping your cruise ship
[07:51:42] <Martin Thomson> in response to Jonathan: obfuscating the base works too, if you don't know what the base identifies
[07:51:54] <Martin Thomson> but your point is well made
[07:52:55] <Jonathan Lennox> Martin: I'd think an attacker would be a lot more likely to be able to figure out the reference than the offset (e.g. your cruise ship example).
[07:52:58] <Martin Thomson> in response to Roger: the baseline and reference are different points, and the language is quite precise
[07:53:24] <Martin Thomson> Jonathan: right, a commonly used reference location is more likely to be known
[07:54:15] <Jonathan Lennox> I'm afraid an obfuscator ignorant of relative location might try to obfuscate the base and pass the relative through verbatim...but probably an obfuscator shouldn't pass unknown elements through regardless.
[07:54:23] <Jonathan Lennox> Did you want your comment to Roger relayed?
[07:54:29] <Martin Thomson> mic: PIDF-LO already has uncertainty, I don't think that Roger's concern is relevant
[07:54:36] <Martin Thomson> that one please
[07:55:12] <Martin Thomson> there's a discussion on obfuscating and supplementary (and unknown) data in my obscuring draft
[07:55:33] <Jonathan Lennox> okay, great
[07:55:38] <Martin Thomson> mic: PIDF-LO has confidence (95%)
[07:55:58] <Martin Thomson> don't bother with that one then
[07:57:03] <Jonathan Lennox> For my ignorance, not to bother the room ... how does the backward-compatibility story work for relative location? I.e. would someone who didn't know about it think the reference point is actually the location being described?
[07:57:26] <Martin Thomson> the backward compatibility story is one that I had to work very hard to get right
[07:58:02] <Martin Thomson> the _baseline_ is the bit that is visible to a legacy user, and that should enclose (with its uncertainty) the relative location
[07:58:22] <Martin Thomson> the _reference_ location adds to the baseline (or replaces it, for geodetic) and is used as the basis for relative location
[07:58:58] <Martin Thomson> Write it up Ray, that would be great
[07:59:41] <Jonathan Lennox> Martin: ah, excellent...sounds good.
[07:59:53] <Jonathan Lennox> Did you want "write it up ray" at the mic?
[08:00:03] <Martin Thomson> Jonathan: (nah, he's in the room)
[08:00:09] <Jonathan Lennox> Ok
[08:00:49] <ray> msg received....
[08:00:58] <Martin Thomson> For Ray: can you express your identifier as a URI?
[08:01:37] <ray> I doubt it - it needs <L3 proto, L4 proto, L4 src_port, L4 src_addr, L4_dst_port, L4 dst_addr>
[08:01:52] <ray> oops - I mean L3_src_addr and L3_dst_addr, not L4
[08:02:07] <Martin Thomson> Right, the complete flow identifier, not just one end of that flow
[08:02:12] <ray> exactly
[08:02:29] <ray> if someone already defined a URI identifier for that I suppose we could re-use it
[08:02:32] <Martin Thomson> these slides aren't on the materials, are they?
[08:03:06] <Martin Thomson> ray: don't try to make your own URI if one doesn't already exist, we tried that once...bad idea
[08:03:12] <ray> the problem with the current extension for ip/port is that they don't lend themselves to consistent answers
[08:03:49] <ray> since if we re-used the current ones you could get different <ip v="x"> protocol versions for src/dst
[08:04:00] <ray> @martin - no, I won't try to define a new one
[08:05:24] <Martin Thomson> mic: can someone ask the ---priv question please?
[08:06:59] <Martin Thomson> mic: in response to Terry, that is excellent feedback - and it would be my feedback too (c.f. ---priv)
[08:07:53] <Martin Thomson> obfuscation isn't going to cut it :)
[08:08:01] <Jonathan Lennox> Reminds me of the Usenet Mapping "ICBM Address" field.
[08:11:45] <ray> sorry - that was me at the mic....
[08:12:21] <Martin Thomson> I can recognise most of the voices now :)
[08:13:25] <ray> I wasn't sure who else is listening to the audio...
[08:15:32] <Martin Thomson> mic: In response to Brian: consider it axiomatic. That is: always assume that location will change. I know very few people/targets that are nailed to the one place.
[08:16:09] <ray> cullens in the queue first
[08:16:29] <Martin Thomson> ooo, I want to hear what Fluffy has to say
[08:19:54] <ray> if I ever start to sound like ekr pls shoot me
[08:20:21] <Martin Thomson> you are plenty clear enough for me. those american's have trouble with non-american accents, that's all.
[08:26:38] <richard.barnes> apparently australian's have trouble with punctuation :)
[08:27:32] <Martin Thomson> sometime's
[08:30:35] <Martin Thomson> mic: in the spirit of naval gazing, a lot of the stuff we built was too complex for real uses, not complex enough for acedemics (read: OGC)
[08:31:18] <Martin Thomson> Hannes hit the nail on the head
[08:32:37] <Martin Thomson> mic: and I should add, it's worse that it *looks* more complicated than it really is as well
[08:33:00] <Jonathan Lennox> Oh, sorry, didn't see that one...
[08:33:10] <Martin Thomson> no problem, it should be OK
[08:33:33] RjS leaves the room
[08:33:35] Randall Gellens leaves the room
[08:33:42] ray leaves the room
[08:34:02] <Martin Thomson> Thanks for the relaying Jonathan
[08:34:02] Jonathan Lennox leaves the room
[08:34:39] suzukisn leaves the room
[08:42:18] Kaveh Ranjbar leaves the room
[08:42:39] RjS joins the room
[08:42:56] RjS leaves the room
[08:44:09] tsuichi leaves the room
[08:46:57] richard.barnes leaves the room
[08:49:36] Martin Thomson leaves the room
[08:52:13] tsuichi joins the room
[08:58:09] tsuichi leaves the room
[09:00:15] suzukisn joins the room
[09:00:43] tsuichi joins the room
[09:02:27] RjS joins the room
[09:06:40] RjS leaves the room
[09:08:43] suzukisn leaves the room
[09:11:55] genki_yasutaka leaves the room
[10:19:41] tsuichi leaves the room
[13:02:53] RjS joins the room
[13:27:45] RjS leaves the room
[22:38:13] RjS joins the room
[22:54:01] RjS leaves the room
[23:11:40] RjS joins the room
[23:12:31] RjS leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!