17:01:20 <isabela> #startmeeting tor network team meeting
17:01:20 <MeetBot> Meeting started Mon May 30 17:01:20 2016 UTC.  The chair is isabela. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:20 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:01:37 <Sebastian> looks like you can ;)
17:01:41 <isabela> Sebastian: !
17:01:42 <isabela> hehe
17:01:57 <isabela> ok folks who wants to start with updates
17:02:13 <armadev> i'll go. i have three interesting things:
17:02:14 <isabela> isis special athena Yawning dgoulet asn armadev Sebastian teor etc
17:02:32 <armadev> a) i did two talks last week on tor hidden services, sort of like the 32c3 talk but more indepth, in eindhoven and then utrecht
17:02:41 <armadev> we'll see if any of the crypto people i talked to comes up with more things
17:03:04 <armadev> b) i've been shepherding a pets paper on an improved bridge distribution strategy. i think it is a good one that we should try. more detail when it goes public, soon hopefully.
17:03:21 <armadev> c) i sent my schedule in more detail, so people can know how to predict me. it is super ugly until sept.
17:03:22 <armadev> eof
17:04:13 <armadev> (..next? :)
17:04:49 <armadev> (oh, re 'a', i realized more things i would want to see in prop#224 and prop#250. i'll try to make a list and then see if they're in the spec yet.)
17:04:49 <Sebastian> I'm here to listen, got no update this week
17:05:15 <Yawning> what
17:05:18 <special> armadev: you should, and feel free to share that list early
17:05:20 <Yawning> oh
17:05:23 <Yawning> right
17:05:32 <isabela> hello Yawning
17:05:34 <Yawning> national holidays don't matter or something
17:05:51 <isabela> you welcome to skip this one
17:06:02 <armadev> we live in the postnational future! but yes, holydays are fine too.
17:06:03 <isabela> i am here cuz i dont care about US holidays
17:06:09 <isabela> hehhe
17:06:36 <Yawning> uh
17:06:41 <Yawning> I did stuff
17:06:54 <Yawning> I should shelf working on basket2 to work on core tor
17:07:10 <Yawning> but I didn't and so basket2 is almost something I am comfortable deploying
17:07:11 <Yawning> w/e
17:07:49 * dgoulet can go
17:07:52 <special> I've done some work on HS descriptors, in particular trying to get the cross-certification stuff nailed down and playing with more unit tests. Many distractions from that work right now
17:09:00 <dgoulet> ok I'll go, seems like anytime is a good time
17:09:07 <dgoulet> Some of my time last week was to recover from physical exhaustion due to conference. Some other part is prop250 fixes and testing. Some "pre-final" branch for review in ticket16943_029_05 that is if asn doesn't find anything to fix :).
17:09:08 <isabela> :)
17:09:11 <dgoulet> Last, some review happened on review-group-2 tickets, seems we are down to 3 needs_review ticket in 029 :).
17:09:18 <dgoulet> altough 211 open ticket in 029! lol
17:09:30 <dgoulet> we now have this issue where we review a lot! but don't address "needs_revision" too much :P
17:09:35 <dgoulet> (maybe it's no problem)
17:10:11 <dgoulet> --
17:10:42 <isabela> ok! about 211 open tickets :) i shared a plan for us to make it smaller and not go crazy
17:11:00 <Yawning> too late on the latter for me
17:11:15 <Yawning> but my sanity was already long gone before I started contributing to tor
17:11:20 <isabela> :)
17:11:55 <isabela> i have an intense month ahead where this will be very important work (making 029 smaller)
17:12:13 <dgoulet> moving ticket out of 029 sounds like a good idea lol
17:12:27 <isis> i read the pets paper that armadev mentioned, worked on a bug with libscrypt not being found on some systems, and then did bridgedb work
17:12:28 <isabela> hope we can make it happen pleae let me know if any questions we can talk more on discussion time
17:13:05 <isis> i also have been preparing for a panel and talking to snowden on wednesday
17:13:14 <isabela> !
17:13:25 <isabela> thats pretty cool :)
17:13:49 <isis> except by "preparing" it is more like "yolo i will mostly wing it because i have never had to publicly have a conversation with other people before"
17:14:19 <armadev> isis: every time i talk to ed, he asks me about padding, and mixing, and basically how to improve tor against traffic confirmation attacks.
17:14:23 <armadev> so, now you know. :)
17:14:39 <isis> armadev: aha, thanks :)
17:14:52 <dgoulet> we all know!
17:14:54 <dgoulet> :P
17:15:01 <isabela> heheh
17:15:48 <isabela> ok, more updates?
17:15:52 <isabela> i think asn is away
17:16:28 <isabela> teor is in the cloud flying?
17:16:39 <armadev> or dead asleep at his 3am
17:16:44 <isabela> hehehe
17:16:46 <dgoulet> deep asleep
17:16:51 <isabela> :)
17:17:03 <isabela> alright, should we go to discussion points then?
17:17:40 <armadev> FAQ for prop#250: Why not use an external randomness source, like stock prices or the head of a blockchain? We should answer this one in the proposal.
17:17:40 <isis> armadev: do those researchers who wrote the PETS bridge distribution paper have code for how they calculated their statistics? (or better, an actual implementation?)
17:17:51 <armadev> isis: yes, they say they have an implementation.
17:18:05 <isis> cool, i would like to see it, if they would like to share
17:18:57 <armadev> Is there a (ticket with) analysis of the complexity of choosing a vanity onion name, in the old style and the new style? I've heard everything from "it's about the same" to "it will become much easier".
17:19:19 <asn> istr there is a section in the end of prop224 about this
17:19:56 <asn> (hey people. popping in for a sec. last week i was doing prop250 stuff and code reviews. this week im going to be busy with prop224/250 stuff and more code reviews)
17:21:17 <armadev> yes, i see appendix c. i will try to learn from dan how much computation is involved in those additions.
17:21:37 <armadev> did somebody working on prop#250 see my FAQ above? or should i open a ticket or mail it :)
17:21:40 <dgoulet> special and teor actually worked on that in montreal with some Yawning around it :)
17:21:41 <Yawning> did you all
17:21:51 <Yawning> change the algorithm I use
17:23:00 <armadev> and then thought #3:
17:23:02 <armadev> Idea: We could encode, using a few bits in the name itself, how many replicas the onion service is publishing to. That way we can scale the HSDir side of things too. Are there other ways to scale the HSDir list for an onion service?
17:23:24 <special> 224 currently says the number of replicas is a consensus parameter.
17:24:08 <Yawning> https://github.com/Yawning/horse25519/blob/master/src/ref10/ge_add.h
17:24:14 <Yawning> if not
17:24:21 <armadev> ok. so putting it in the name will let us choose per name. and in the consensus will be a global param. ok.
17:24:49 <special> I would be interested to hear the use cases for choosing it per name
17:24:52 <armadev> yawning: can you compare those additions to the cost of, say, a sha1?
17:25:08 <armadev> special: if you're facebook and you know that your six hsdirs are going to be sad sad sad
17:25:37 <armadev> you can scale your intro points yourself, and do various tricks like tvdw's proposal to scale rendezvous responses
17:25:40 <armadev> but the hsdirs are out of your hands
17:26:25 <Yawning> armadev: "depends"
17:26:29 <Yawning> on how fast your sha is
17:27:32 <armadev> then please fork and give me all answers? :) unless there are a lot? i most want to know about the fast sha1's -- the ones on facebook's cluster for example.
17:29:51 * isabela wonders if folks had the opportunity to read her email and if they have questions
17:29:54 <Yawning> for most systems the add is faster than hasing
17:30:10 <Yawning> I say most because some architectures have hardware sha
17:32:11 <armadev> the thing i'm actually comparing to is two operations. stick with a given pq=n, increment e, compute sha1 of resulting key.
17:32:26 <armadev> i guess that increment e step is basically free compared to the sha1?
17:33:08 <armadev> but yes, i think we should compare "the fastest relatively-easy-to-get hardware" for both cases.
17:33:20 <armadev> and, how much faster
17:33:35 <armadev> dgoulet: re isabela's mail, are you or asn or me the right one to answer about the sponsorR 029 tickets?
17:33:49 * dgoulet reading email
17:34:29 <dgoulet> armadev, isabela: oh we just talked about it before this meeting and seems we have all the tickets on trac for SponsorR (except one that I need to open)
17:36:29 <armadev> did we decide between 029 and 030, there?
17:36:35 <isabela> not yet
17:37:10 <isabela> dgoulet: great - please just make sure owners are assigned, points etc
17:37:26 <dgoulet> isabela: I'll review them all thoroughly but I'm confident we have a good state right now, i'll let you know once done
17:37:27 <isabela> remember ppl will be travelling and so on
17:37:37 <isabela> cool thanks a lot!
17:38:57 <armadev> ok i'm going to disappear for a bit. please let me know if there's something i should be useful on for you this week.
17:39:07 <dgoulet> should we create a 0.3.0 milestone in trac or 030 has been decided and 0.2.10 might be the one?
17:39:17 <armadev> i just made up 030 right there
17:39:23 <armadev> but 0.2.10 sure has a lot of digits
17:39:58 <isabela> i think we can ping nick on it when he is back
17:40:02 <isabela> maybe tomorrow
17:40:05 <dgoulet> ack
17:40:06 <isabela> or in the list as well
17:40:22 <isabela> i dont really remembe if he has a preference here
17:40:33 <isabela> are we good with this meeting?
17:40:37 <isabela> should I stop the bot?
17:40:50 * dgoulet is good
17:41:03 * isis also
17:41:31 <isabela> !
17:41:36 <isabela> #stopmeeting
17:41:40 <isabela> blah
17:41:47 <isabela> #endmeeting