17:00:28 <nickm> #startmeeting weekly network team meeting
17:00:28 <MeetBot> Meeting started Mon Jan 30 17:00:28 2017 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:28 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:00:35 <nickm> hi hi!
17:00:55 <nickm> status updates to begin!
17:01:33 <asn> hello :)
17:01:38 <dgoulet> hi!
17:01:39 <nickm> Last week I merged a bunch of things in advance of the feature freeze, and helped interview a bunch of candidates, and tried to stay up-to-date on things
17:01:44 <isabela> buenas
17:01:50 <nickm> we're now at 30 tickets left in 030
17:02:31 <nickm> this coming week I'm going to try to continue helping with our hire process, try to help onboarding ahf, and try to fix as many 030 bugs as I can.
17:02:44 <nickm> I hope I can get to some sponsor4 early-stage measurement
17:03:41 <nickm> and so yeah!
17:03:43 <nickm> who's next?
17:03:45 <nickm> (hi all)
17:03:46 <asn> i can go!
17:04:05 <asn> Hello. During past week I did various reviews on HS/guard/fuzzing-related
17:04:05 <asn> branches. I also did lots of design work for prop224:
17:04:05 <asn> https://lists.torproject.org/pipermail/tor-dev/2017-January/011816.html
17:04:05 <asn> ohttps://lists.torproject.org/pipermail/tor-dev/2017-January/011841.html I also
17:04:05 <asn> listened in multiple interviews for hiring in network team.  This week I will
17:04:08 <asn> wrap up the prop224 torspec stuff, and start work on some 0.3.1 prop224 stuff
17:04:10 <asn> (#21334) so that it's ready for early merge in 0.3.1. ALso looking forward to
17:04:13 <asn> ahf onboarding! EOF. Next?
17:04:17 <ahf> hello!
17:04:28 <asn> \o/
17:04:29 <nickm> hello ahf!
17:04:31 <dgoulet> wild ahf appears!|
17:04:34 <ahf> :-D
17:04:46 <isabela> yo!
17:04:53 <ahf> thanks for all the replies to my network-team mail - that have been very useful for me
17:05:26 <ahf> i plan on joining in on wednesday at 13 UTC (12 local time) and say hi. what should i begin with? start on 030 bugs?
17:06:45 <nickm> ahf: good question.  if you see any 030 bugs that you can make progress on, that would be helpful.  But they might be kinda specialized by this point...
17:07:04 <nickm> The sponsor4 measurement stuff might be a good place to start too
17:07:07 <nickm> isabela: what do you think?
17:07:11 <ahf> yes, i shouldn't be blocking anything that have to go out of the door if you guys are much faster there.
17:07:26 <ahf> cool. i can find some information on that on the trac wiki?
17:07:48 <isabela> yes!
17:07:53 <nickm> https://trac.torproject.org/projects/tor/wiki/org/teams/NetworkTeam/Sponsor4Plan yes
17:08:11 <nickm> #21205 and #21209  are the early-stage "measure and design" tickets
17:08:25 <nickm> most of the measurement stuff ought to be simple; if it's hard, we're probably doing it wrong :)
17:09:20 <isabela> nickm: we have sponsor4 in the menu now btw :) i will add it to the tickets your tagged with the sponsor4 keyword
17:09:31 <ahf> oh, looks cool. think this was also brought up by some people at the tor relay operators meeting at 33c3 - at least sending diff's back and forward
17:10:05 <nickm> yeah -- and there are other possibilities too
17:10:09 <nickm> like improved compression etc
17:10:11 <ahf> yep
17:10:20 <nickm> and changing how we choose which mds to download
17:10:43 <ahf> what does mds mean here?
17:10:48 <nickm> microdescriptors
17:10:51 <ahf> ah, of course
17:11:14 <nickm> one reason that downloads are bigger than they need to be is that the client says which microdescriptors it wants, and the directory cache sticks them all together and zlib-compresses them
17:11:26 <nickm> so the request is big and the response can't be pre-compressed
17:11:55 <nickm> but in theory, the client might be able to say "tell me all microdescriptors that appear in consensus X but not in consensus Y"
17:12:04 <nickm> or something like that
17:12:13 <ahf> yep.
17:12:34 <nickm> ahf: also please don't be shy about asking questions: we're on IRC and we love to talk about code :)
17:12:46 <ahf> trust me, i'm not shy with asking questions :-)
17:13:10 <ahf> and also talk about the code. i remember writing the code for the dir_client in talla for this, so i have an idea how this works
17:13:29 <ahf> and i do remember some of this code from the C tor codebase as well when toying around with adding some debug information here and there
17:13:50 <ahf> i'll look at the two bugs that are related to and get an overview of it and then maybe we should chat when you check in on wednesday?
17:13:55 <ahf> i assume it's you and me that will be working on this?
17:14:00 <nickm> sounds good to me!
17:14:04 <nickm> also possibly the new hire too
17:14:40 <ahf> yep! asn told me about the interviews the other day and how it went - looking forward to hear who is going to get selected :-)
17:14:54 <nickm> me too!
17:15:17 * dgoulet goes for quick status report
17:15:21 <ahf> let's chat on wednesday when you check in - i will have read all the info you just linked to be then and we could get the ball rolling
17:15:24 <ahf> cheers!
17:15:29 <nickm> peace!
17:15:31 <dgoulet> status: Code review for 030 tickets. Fixed some opened/needs_revision and triage more tickets out of 030. Worked a bit on prop224 #20657 and thanks to asn now we have the hs circuitmap working for services (pulled in my dev. branch).
17:15:34 <dgoulet> So many interviews last week also! and last one today! I'll have to make progress a bit more with prop224 code this week so expect me to do a bit more of that than my usual ticket work.
17:15:37 <dgoulet> --
17:15:52 <dgoulet> (btw we had 2 good threads about prop224 on tor-dev@ so progress++)
17:15:54 <nickm> asn, dgoulet: quick question for the prop224 stuff, where maybe you already answered it...
17:16:02 <asn> aha
17:16:08 <nickm> did you remember for the directory stuff that we want to support offline master secret keys?
17:16:32 <dgoulet> offline keys for "directory stuff"?
17:16:38 <dgoulet> nickm: you mean offline keys for the service?
17:16:47 <nickm> yeah.  The service should be able to keep its master key offline
17:16:58 <nickm> (not mandatory, but optional)
17:17:19 <dgoulet> yes we do have that in mind but so far considered a "prop224-extra" that is not in the critical path of a the viable solution we want to release
17:17:33 <nickm> Sure; doesn't need to be in first release...
17:17:39 <asn> the directory code should not block this design.
17:17:46 <asn> in the sense that we are signing the descriptor with the blinded key
17:17:56 <asn> and not the long-term master key (that could be offline in theory)
17:18:07 <nickm> well hang on
17:18:20 <nickm> I thought we were signing the descriptor with a different key, and signing _that_ key with the blinded key
17:18:28 <dgoulet> yeah there is a desc. signing key
17:18:31 <nickm> otherwise you need to keep the blinded key online
17:18:31 <dgoulet> created from the blinded key
17:18:38 <asn> ^^^
17:18:46 <dgoulet> In this design, a hidden service's secret identity key may be
17:18:47 <dgoulet> stored offline.  It's used only to generate blinded signing keys,
17:18:47 <dgoulet> which are used to sign descriptor signing keys.
17:19:08 <dgoulet> (yeah sorry made from master public key)
17:19:47 <asn> yep. the ephemeral desc signing key is in the outer layer of the desc as:
17:19:47 <asn> "descriptor-signing-key-cert" NL certificate NL
17:19:49 <nickm> ok. so no need to keep identity secret key or blinded identity secret key online then, since descriptor-signing key is the only one you need while the service is running. great
17:21:04 <dgoulet> nickm: I guess you meant "offline" but yes
17:21:31 <nickm> ok
17:22:12 <nickm> so, any more updates for today?
17:23:04 <nickm> ok!
17:23:13 <nickm> anybody have discussion topics?  I have a couple...
17:23:27 <asn> not me
17:23:43 <isabela> if we had time we could start talking about team meetings in amsterdam
17:23:43 <dgoulet> I do but probably same as yours nickm :)
17:23:59 <dgoulet> isabela: yes that and the freeze/release/branching git situation :)
17:24:00 <nickm> 1 - can we do a quick triage on the 030 tickets  in state "new" and try to assign/defer?
17:24:44 <nickm> 2 - how and when do we decide our feature set for 031 and do ticket triage for it?
17:24:45 <isabela> dgoulet: cool - i would also prioritize the release stuff (we will have more time to talk about team meetings)
17:25:00 <nickm> other topics?
17:25:55 <dgoulet> sounds plenty for the 30 min we have :D
17:26:16 <nickm> ok.  so I have most of the remaining assigned/accepted tickets.
17:26:21 <nickm> and that's okay.
17:26:55 <nickm> but for the new ones, I'd like help
17:27:06 <nickm> maybe we can find some for ahf too
17:27:13 <nickm> there are 9 of them
17:27:20 <dgoulet> this is #21297 super uncertain ... #21302 came out of it so I guess we could put it in tor unspecified.... ?
17:27:41 <asn> dgoulet: agreed
17:27:42 <nickm> ok
17:27:47 <asn> dgoulet: we dont seem close to figuring it out right now
17:27:53 <dgoulet> yah..
17:28:16 <dgoulet> boom
17:28:39 <nickm> #21294 is a documentation fix that should be easy
17:28:49 <nickm> #21290 is a little easy autoconf and documentation stuff
17:29:03 <dgoulet> oh teor's ticket #21283 seems it should be in needs review
17:29:13 <nickm> #21266 -- can I make that "assigned" for you, dgoulet ? you're already "owner" on it...
17:29:23 <dgoulet> nickm: sure!
17:30:47 <nickm> #21074 is a "how do I reproduce this" issue.  I'd say "unspecified, needs_info"
17:30:52 <dgoulet> (I'm still quite confused with the difference of Accepted and Assigned because both set your nick as the Owner)..
17:31:10 <nickm> Assigned you can use to set other people's nicks
17:31:19 <nickm> but I don't know what the difference is either :)
17:31:27 <dgoulet> ahah ok
17:31:31 <dgoulet> let,s consider it the same :)
17:33:00 <dgoulet> nickm: is this an easy fix for you? #21108
17:33:09 <nickm> I'm not sure
17:33:13 <dgoulet> asking because seems you looked at it already
17:33:15 <nickm> I guess I could take it on -- assign it to me?
17:33:30 <dgoulet> there
17:33:45 <nickm> #21044 looks pretty complicated, and it's in an area where any changes we make are likely to destabilize something
17:33:50 <nickm> so maybe deferring would be smart...
17:34:14 <dgoulet> indeed 031
17:34:15 <asn> sounds good
17:34:59 <dgoulet> can we shamelessly give some to ahf? AHAH :D
17:35:03 <nickm> IIRC #21061 is a user with a setup that's different from ours, who fixed their issue IIUC
17:35:14 <nickm> dgoulet: like which ? :)
17:35:19 <dgoulet> #21242
17:35:34 <dgoulet> nickm: ok let's close the ticket (#21061)
17:36:05 <nickm> ok
17:36:08 <nickm> what's left?
17:36:32 <dgoulet> ok taking #21294, easy man page fix
17:37:02 <asn> i bet ahf could handle #21266.
17:37:18 <dgoulet> asn: indeed
17:37:23 <nickm> #21242 -- I haven't been able to reproduce it, though maybe the answer is to stare at the assertion
17:37:32 <nickm> I suspect that the prop271 code is behind 21242
17:38:18 <dgoulet> nickm: that #21290 seems quite easy to fix also but we need a name and naming is _hard_ :D ;)
17:39:19 <nickm> I'll take on #21242 I guess
17:39:26 <asn> damn #21242
17:39:32 <asn> i hadn't seen this one.
17:39:36 <nickm> asn: unless you want to look at it? ;)
17:39:38 <asn> nickm: i can take a look at it tomorrow
17:39:45 <nickm> asn: ok, very grateful.  if you get stuck let me know
17:39:48 <asn> ack
17:40:32 <nickm> dgoulet: wrt 21266 -- or maybe you could add the log checks to one of the test cases, as an example for what we have in mind, and ahf could go over the rest
17:41:04 <dgoulet> nickm: oh we have a lot of examples in the HS unit test for this so I,m sure it's pretty easy for ahf to pick it up :)
17:41:12 <nickm> hm, ok
17:41:14 <dgoulet> I'm confident
17:41:22 <nickm> great
17:41:54 <nickm> ahf: I think we just assigned you an 030 unit-test-improvement ticket
17:42:23 <dgoulet> nickm: 031 ? or you really want that in 030? :) #20895
17:42:24 <nickm> dgoulet: could you explain what we mean on the ticket, though?  Or maybe just quote the review?
17:42:33 <nickm> dgoulet: let's postpone 20895
17:42:41 <dgoulet> ok moving it
17:42:44 <nickm> thanks!
17:42:59 <nickm> 21290 is the remaining "new" one.
17:43:17 <dgoulet> yeah which is the hardest part is finding a good replacement
17:43:40 <nickm> I think --enable-fragile-hardening might be right
17:44:14 <asn> i can provide a summary in #21266 for ahf
17:44:20 <nickm> asn: many thanks!
17:44:21 <dgoulet> nickm: does sound good
17:45:11 <nickm> ok.  anybody want to pick up #21290 or shall we wait for somebody to have time
17:45:24 <dgoulet> nickm: I can take it
17:45:36 <dgoulet> boom
17:45:48 <nickm> yaaaay!
17:45:58 <nickm> so, nothing "new", down to 25 tickets
17:46:27 <nickm> I think we should consider needs_iknformation and needs_revision for triage but maybe not right now since we've got that other thing in 14 minutes :)
17:46:35 <dgoulet> yah eheh
17:46:44 <nickm> so, Q2: any thoughts on what our list of goals for 031 should be
17:47:13 <dgoulet> prop224, mikeperry's padding at least?
17:47:40 <nickm> yes, and some sponsor4 measurement stuff (ideally w prelim fixes)
17:47:41 <dgoulet> #16861
17:48:08 <nickm> actually, I just moved all the 030 needs_revision stuff to 031
17:48:55 <nickm> maybe we should also set some refactoring/testing goals
17:49:09 <nickm> and i wonder if we should do groundwork for big create cells
17:49:18 <nickm> (if we're hoping to do pq)
17:49:34 <nickm> maybe we should go over the proposals and build a possible higher-level list of goals, and then see what we realistically have time for
17:49:45 <dgoulet> sounds wise yes
17:50:00 <dgoulet> prop224 is still a long way, lots of work still need to be done and will add a lot of code
17:50:09 <dgoulet> 0.3.2 will be a fun "let's stabilize it" version
17:50:30 <dgoulet> I still have this dream that we actually make a version with NO feature, just fixes and stability cleaning up our Trac in the process :D
17:50:40 <nickm> hee
17:50:42 <nickm> could be neat
17:51:04 <nickm> for deciding possible featurs for 031, how about a wiki page?
17:51:15 <nickm> s/features/high-level goals/
17:51:26 <dgoulet> nickm: and we add our propositions in there?
17:51:31 <asn> i think that's a good idea
17:51:47 <asn> i mean, perhaps each release deserves a wiki page anyway.
17:51:52 <nickm> yeah
17:51:58 <nickm> maybe we have one and forgot :)
17:52:15 <asn> (Added info to #21266)
17:52:33 <nickm> if anybody has time to start that page today, great.  Otherwise I should be able to get to it in like 5-24 hours?
17:52:58 <asn> i doubt i can do it today.
17:53:20 <nickm> ok, i'll make a note
17:53:31 <nickm> and I'll also call this meeting done, barring last-minute things, so we can go do that other thing!
17:54:02 <isabela> we have a release guidelines page
17:54:05 <asn> thanks. good meeting!
17:54:38 <nickm> #endmeeting