16:59:57 <nickm> #startmeeting weekly network team meeting.
16:59:57 <MeetBot> Meeting started Mon Oct 31 16:59:57 2016 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:59:57 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:00:04 <nickm> welcome, everybody, to the network team meeting.
17:00:10 <nickm> Also, happy halloween (if that's a thing for you!)
17:00:14 <armadev> douglas_: but, do they use it in the transparent proxy way, which probably shows poor judgment?
17:00:37 <nickm> dgoulet: armadev: asn: athena: yawning: isis: chelseakomlo: isabela: welcome!
17:00:45 <dgoulet> hi!
17:00:54 <nickm> as usualy, we'll start with status updates, then discussion.
17:00:59 <nickm> hi, dgoulet!
17:01:02 <nickm> who else is here?
17:01:10 <armadev> i!
17:01:17 <Sebastian> Hi there. I have to leave abruptly soon I'd like to share a couple of things
17:01:56 <nickm> hi armadev, hi Sebastian !
17:02:01 <nickm> Sebastian: you should probably go first then
17:02:08 <nickm> (isabela said she might have to leave early too)
17:02:15 <Sebastian> I kinda restarted the work on removing naming from Tor last week, discussed lots about version recommendations and how that fits into making releases for stuff. And I guess my plan for next week is to actually make some mergable stuff for the naming removal
17:02:24 <Sebastian> I'm already done :)
17:02:36 <Sebastian> Wonder what I forgot ;)
17:02:39 <nickm> done with your status update or done with the naming?
17:02:40 <chelseakomlo> nickm: hi! :)
17:02:48 <nickm> hi chelsea!
17:02:48 <Sebastian> status update!
17:02:55 <armadev> and my status update is easy: i have been messing with #19969 and noticing #20499 and i filed a few new tickets like #20501 and #20502 and #20469
17:02:59 <nickm> Sebastian: okay!  Sounds good!
17:03:03 <Sebastian> naming i didn't really get a chance to dig in
17:03:11 <nickm> armadev: if you help us decide what to (not) backport to 028 or earlier, that would rock.
17:03:17 <nickm> armadev: we're still noodling at tor-lts issues.
17:03:31 <chelseakomlo> i have a short update: I'm going to finsih #18873 with the feedback from dgoulet, and then I have a proof of concept for running tor integration tests in a docker container
17:03:59 <nickm> neat!
17:04:09 <nickm> what will the benefit from that be?
17:04:17 <armadev> (right. one of my conclusions from the last lts discussion was that we need a plan for how an old update will get into the hands of users (e.g. via a debian package), or it's less clear that doing the update will matter.)
17:04:36 <nickm> armadev: I believe debian packages was the plan.
17:04:45 <nickm> armadev: not finding anything else maintained on the same schedule that we care about.
17:05:05 <Sebastian> data point: opensuse is on 0.2.7 apparently
17:05:18 <Sebastian> (and applied the patch we released recently)
17:06:01 <nickm> Sebastian: what would be useful to know is how often opensuse releases come out, and how long they would require Tor to be stable for.  Having it on https://pad.riseup.net/p/tor-lts would rock.
17:06:16 <chelseakomlo> running in a docker container is partly a proof of concept for #20458 but also hopefully to get tor integration tests in the CI
17:06:29 <Sebastian> k, I'll see if I can find out and put it there
17:06:29 <nickm> sounds neat
17:06:34 <chelseakomlo> it should definitely be useful for CI
17:06:43 <armadev> more testing good
17:07:24 <nickm> dgoulet: would you like to go, or shall I?
17:07:29 <dgoulet> I can go!
17:07:37 <nickm> woohoo
17:07:39 <dgoulet> Update: so I have this thing where I'm having trouble remembering what I did last week as I failed to note it down as I usually do... I recall some ticket review from the review-group-11 I believe. Some emails, test network business and unrelated little-t tor stuff.
17:07:56 <dgoulet> but I need to refocus on continuing the prop224 for code where asn is awaiting multiple reviews
17:07:58 <nickm> Sebastian: (my rationale is that we want to know how long we need to support releases for, and how long we need to support current releases for)
17:08:07 <dgoulet> so my plan this week is that priority no1
17:08:36 <dgoulet> nickm: discussion item also, I would like to know if you plan to re-review #17238 soon-ish and if not, I'll move to focus on some other part of prop224
17:08:48 <armadev> makes sense to me -- if we let up on prop224 stuff, we're going to wake up one day and realize it's next summer and we're not done
17:08:52 <nickm> I hope to re-review that this week
17:09:08 <dgoulet> neat
17:09:53 <dgoulet> right prop224 stuff has a lot of code but we need to work on merging it upstream so we can move forward to more bigger tasks that depends on a lot of that code and so forth, we have two big pieces right now "almost ready"
17:09:58 <dgoulet> thanks
17:10:02 <nickm> me now?  or anybody else here?
17:11:07 <nickm> ok, me!
17:11:08 <asn> ugh sorry about that. nickserv kicked me off
17:11:14 <nickm> hi asn!
17:11:58 <nickm> last week I finished the module documentation deliverables for sponsor U.  Now every module that we're not planning to totally revise has documentation about its behavior in a nice doxygen comment at the head of it
17:12:18 <nickm> I worked more on profiling, LTS policy, bugfixes, and reviewed and merged a pile of code
17:12:22 <nickm> I also hacked more on 15056.
17:12:34 <nickm> This week: finish 15056 or bust, since I really need to shift over to guards.
17:12:38 <dgoulet> (#15056)
17:12:39 <nickm> #15056
17:12:41 <dgoulet> :)
17:13:00 <nickm> I also hope to review and merge some of the bug tickets that are pending this week.
17:13:15 <nickm> and figure out bugfixes for 029 etc when I'm blocked.
17:13:28 <nickm> I'm hoping for help on a few of these; I'll save that for discussion time, though.
17:13:49 <nickm> that's it for me.
17:14:17 <nickm> asn next?  then, anybody else?
17:14:44 <nickm> (isis, athena, isabela ...?)
17:14:45 <asn> Hello. This week I did a few reviews, then I revamped my ESTABLISH_INTRO branch
17:14:45 <asn> (#19043) and the prop224 trunnel branch (#20004). This upcoming week I'm
17:14:45 <asn> planning to do more reviews, and write up a torspec branch for the client auth.
17:16:59 <asn> EOF
17:17:03 <nickm> asn: (is #20004 for merging into master or into some prop224 branch?)
17:17:08 <asn> merging into master
17:17:12 <nickm> sounds good.
17:17:25 <asn> it's like a preparation step for prop224.
17:17:25 <nickm> once the torspec branch is done, then should I look at client auth stuff?
17:17:33 <asn> correct
17:17:40 <asn> i still have not gotten to it. too many things happening.
17:17:45 <asn> dont worry i will ping you
17:17:50 <nickm> thanks, asn!
17:17:54 <asn> (currently reviewing #20262)
17:18:22 <nickm> okay!  On to discussion!  But if there's anybody else with an update, please feel free to jump in.
17:18:39 <armadev> asn: if there are bite-sized things i can help with for hs stuff, let me know. i seem to be bad at following through on large things but want to still be helpful.
17:18:48 <nickm> The topics I have are: getting 029 out; 028 backport policy; LTS planning.
17:18:55 <nickm> do we have others in mind?
17:19:26 <dgoulet> small thing:
17:19:43 <dgoulet> about the "0.2.???" that is not really accurate anymore, should we change it or not
17:20:05 <nickm> milestone handling.  got it!
17:20:34 <asn> armadev: perhaps reviewing the prop224 client auth torspec branch when it's out
17:20:35 <nickm> so, for 029, one thing that would be helpful is to make sure that the 029 milestone contains everything that we should fix before 029 is stable.
17:20:47 <asn> armadev: there is also a mailing list thread but this might be way too much noise for you
17:20:55 <asn> armadev: i can ping you and nickm when the torspec patch is ready
17:20:56 <armadev> another topic we might cover: are the weekly bug triage rotation ideas working? i notice asn and dgoulet and asn did it, and this week nobody is.
17:21:01 <nickm> Also fixing and testing 029 bugs, but just making sure we have the list of bugs to fix would be a great help.
17:21:07 <nickm> armadev: added to topics.
17:22:26 <armadev> nickm: re 029 tickets saying 029.. which ones are you worried aren't saying 029 now? the ones that haven't been filed yet? or ones that were misfiled?
17:22:50 <nickm> armadev: both!
17:23:51 <nickm> I guess I can look at everything in 028, 029, 030, " ", etc...
17:24:04 <nickm> focusing on anything with 029-backport marked, or an 029 version.
17:24:06 <armadev> asking people to look through the deferred tickets is a recipe for finding ones to move to 029 that don't reeeally need to move
17:24:19 <nickm> armadev: got a better recipe?
17:24:30 <nickm> armadev: missing stuff that does need to go into 029 is pretty bad too
17:25:09 <armadev> makes sense. i thought we did a triage not too long ago where we dumped 029 stuff out of 029.
17:25:14 <armadev> i guess we didn't do the reverse
17:25:44 <nickm> #19969 is the only 0.3.0 ticket in 029-backport
17:25:59 <nickm> We have multiple candidate fixes; I'm moving it.
17:26:31 <armadev> ok
17:26:49 <dgoulet> hrm... when in doubt actually, I should have put the "029-backport" and then we could triage them instead of trying to find the 030 that we thought it could be a good idea...
17:26:50 <nickm> we have 21 tickets in core tor without a milestone.
17:27:00 <nickm> dgoulet: that would be great; would you like to do that/
17:27:00 <nickm> ?
17:27:09 <dgoulet> nickm: I'm going over the one I triaged lately
17:27:18 <dgoulet> there is at least one I can think of we thought of backporting
17:27:27 <dgoulet> (the cmux ewma issue)
17:28:02 <nickm> oh, the scheduler one?  I'm going to suggest "no" on that one; the consequences are only hypothesized to be good.
17:28:25 <dgoulet> nickm: ack
17:28:25 <nickm> at worst, it just means "the scheduler did nothing"
17:28:30 <dgoulet> basically
17:28:48 <chelseakomlo> i have to sign off, but feel free to send small issues my way, particularly refactoring issues for the moment. 20458 will be something i'll do steadily & will likely take more conversations with teor
17:28:49 <dgoulet> pastly wanted to run some larger shadow network with and without it to see
17:28:49 <nickm> better to test it in some alphas than to find out we've accidentally pessimized when we meant to optimize
17:28:56 <dgoulet> chelseakomlo: o/
17:28:59 <nickm> chelseakomlo: ok cool.  thanks very much!
17:29:14 <chelseakomlo> nickm: more this week :) thanks
17:29:15 <nickm> dgoulet: I'll go over stuff without a milestone and stuff in 0.2.9 again.
17:29:20 <dgoulet> ok
17:29:30 <dgoulet> (I'll do that after the meeting)
17:29:35 <pastly> dgoulet: they're still going. Other changes to the code are making Tor crash. Think I might have finally squashed them.
17:29:53 <dgoulet> ahha ok thanks! :)
17:30:06 <armadev> i am fine leaving the cmux ewma issue unbackported. as i understand it, i agree with nickm
17:30:10 <nickm> So, for 028 and earlier, I suggest that we try to figure out a slightly more formal process for 028 backports, and a much more formal process for 024..027 backports.
17:30:42 <nickm> And I think maybe some kind of milestone refactoring would be appropriate here.
17:30:51 <armadev> for 028 and earlier backports, i remember writing out a proposed backport policy. and then asking isa to mail it to me from her backlog. and i think she did, but now i can't find it.
17:31:05 <nickm> I think maybe I did?
17:31:07 <nickm> anyways.
17:31:25 <nickm> I think we want one policy for 028 ("latest stable") and one for everything earlier
17:31:39 <nickm> and I think we want to have some thing where multiple people sign off on backports.
17:31:51 <armadev> so far so good
17:32:00 <armadev> (ah ha! you did email to me. found it. thanks)
17:32:13 <nickm> and we might want to re-create new branches for backports to 0.2.7 and earlier, since I think we have backported stuff to git that should not have been backported.
17:32:30 <nickm> I think armadev and weasel would be logical backport gatekeepers for <= 027.
17:32:56 <armadev> yes, sebastian and i spoke about release-0.2.7 and how to handle that
17:33:00 <armadev> we didn't come to a conclusion though
17:33:03 <nickm> Though I can't make anybody do it, and I'd love other volunteers.
17:33:13 <armadev> (either back out a bunch of commits to release-0.2.7, or throw it out and try again)
17:33:32 <nickm> let's figure out the approval process, and save the git wrangling for later?
17:33:46 <armadev> ok
17:33:47 <nickm> (I think "make new branches, call the old ones defunct, delete NOTHING" is the best option)
17:34:03 <armadev> release-0.2.7-this-time-for-sure
17:34:43 <armadev> one useful step in 027 land would be to have a list of stuff we already backported. right now i can try to piece that together from git, but i might get it wrong.
17:34:52 <armadev> but, approval process. your proposal sounds plausible to me.
17:35:11 <armadev> especially because it doesn't stop anybody from doing a proposed list of things to backport
17:35:56 <nickm> ok.  So if I'm driving this, nothing will happen this week.  Should we look for somebody else to drive, or assume that doing nothing on <=027 this week is fine?
17:36:04 <nickm> either is ok w me
17:36:54 <armadev> i don't think the world will end if it waits
17:37:01 <nickm> ok
17:37:36 <armadev> speaking of which, what was the reasoning for not doing an 0.2.7.7 with the trove bug fix?
17:37:40 <nickm> For 028, I was planning to continue to use my judgment on what to backport. Any objections?
17:37:54 <armadev> was it because release-0.2.7 already had a bunch of stuff in it and solving that at the time didn't seem worthwhile?
17:38:01 <nickm> armadev: adding that to the list of discussion topics.
17:38:07 <armadev> ok :)
17:38:42 <nickm> assuming no objections then. :)
17:38:43 <armadev> i guess one difference between 028 and 027-and-earlier is that tor browser still uses 028
17:38:52 <nickm> next topic: milestone refactoring.
17:39:02 <dgoulet> nickm: fine by me for 028
17:39:03 <armadev> so if we stick stuff in an 028 update, then in theory actual tor users -- ours -- can get it on the next tor browser release
17:39:17 <dgoulet> as for 027, finding gatekeepers is correct first step I believe and then they can drive it
17:39:20 <nickm> dgoulet raised this as 0.2.???, but I'd like to propose this scheme.
17:39:29 <armadev> ergo, let me put in a weak vote for paying attention to the tor browser release schedule
17:39:44 <nickm> dgoulet: got any gatekeeper nominations?
17:39:46 <nickm> :)
17:39:57 <armadev> dgoulet: well, wait, why do gatekeepers necessarily need to drive it
17:40:28 <armadev> an equally valid role for them would be to be oracles that say 'yes' or 'no' when asked about a given patch
17:40:37 <armadev> which then leaves a hole for who is driving
17:40:45 <dgoulet> armadev: well at some point we need some people to actually do something and if the gatekeepers job is to gatekeep <= 027 bugs, it's kind of implicit that they might want to drive the discussion on the process they want?
17:41:14 <dgoulet> and then once that process is established, they can indeed at that point just sign off
17:42:03 <dgoulet> nickm: to answer your question, kind of uncomfortable putting people on the spot :P
17:42:17 <dgoulet> we can do an open call on network team mailing list I guess?
17:42:27 <nickm> sure, and maybe ask packagers.
17:42:29 <nickm> too
17:42:31 <dgoulet> yup
17:42:33 <armadev> dgoulet: in theory yes. in practice if weasel and i are the gatekeepers, we will not also be the backport managers
17:42:57 <dgoulet> armadev: right, deciding the process and rules is more what I had in mind
17:43:07 <dgoulet> not the "next 2 years of gatekeeping" :)
17:44:37 <nickm> ok. on to milestones? :)
17:44:46 <armadev> sounds good
17:44:57 <nickm> hm. 15 min left.
17:45:29 <nickm> so, I'm going to suggest that once a release is stable, we no longer use an "Tor: x.y.z" milestone for it.
17:45:53 <armadev> gosh!
17:45:53 <nickm> Instead, we have a "Tor: stable" milestone for bugs in the supported stable release...
17:46:14 <nickm> and a "Tor: lts" milestone for bugs to be backported to the LTS releases (or not)
17:46:41 <armadev> would we still do the keywords things we do now?
17:46:50 <nickm> and that we rename "Tor: 0.2.???" to "Tor: later release"
17:47:00 <nickm> armadev: question too vague; try again
17:47:13 <armadev> would we still use keywords like 028-backport and 027-backport
17:47:52 <isis> ugh, sorry, i forgot about the time difference
17:48:08 * isis reads backlog
17:48:31 <isis> fwiw, i've been working on implementing and writing down the crypto for the rbridge thing
17:48:46 <nickm> armadev: I don't know!
17:48:48 <nickm> armadev: maybe!
17:48:52 <nickm> probably even.
17:49:00 <nickm> isis: cool!
17:49:14 <nickm> isis: are you targetting deployment or prototype ?
17:49:22 <isis> so not really in network-team-land, since i need to finish up this otf stuff
17:49:27 <nickm> ack
17:49:32 <dgoulet> nickm: so hrm, we have all 030 ticket to "Tor: Stable" milestone, then if we want to backport one, once merged to upstream we set it to "Tor: LTS" with version number in the keywords?
17:49:39 <isis> nickm: i think i can actually make something production ready
17:49:40 <dgoulet> something like that would be the process?
17:50:04 <isis> nickm: there's no pairings, it's reasonably fast, it's in rust
17:50:16 <nickm> dgoulet: so right now, if there's a bug that's only in 0.3.0, it belongs in tor: 0.3.0.x-final
17:50:24 <dgoulet> yes
17:50:26 <nickm> dgoulet: so right now, if there's a bug that's only in 0.2.9, it belongs in tor: 0.2.9.x-final
17:50:35 <nickm> and that wouldn't change until 0.2.9 becomes stable.
17:50:53 <nickm> once 0.2.9 is stable, we close tor: 0.2.9.x-final
17:51:05 <nickm> and bugs that are about 0.2.9.x go into "Tor: stable".
17:51:26 <nickm> isis: cool!  please tell me how you're liking rust some time
17:51:46 <nickm> no need to decide on the milestone stuff today; just wanted to float the idea for comment.
17:51:52 <armadev> are we going to be sad later that some tickets with milestone Tor: stable went into one stable, and others went into a different stable?
17:52:00 <armadev> (or will we be happy later)
17:52:47 <armadev> at least in theory, right now we can ask trac for a list of tickets that went into a given stable tree (e.g. 0.2.8.x-final)
17:53:05 <nickm> hm.  that is a good point.
17:53:14 <nickm> counterpoint: please look at tor: 0.2.[4567].
17:53:27 <nickm> FWICT they are deeply unknown stuff.
17:53:35 <nickm> so maybe we should think more about this.
17:53:41 <armadev> i think right now the milestone is the highest-numbered branch it went into,
17:53:51 <nickm> 7 minutes left.  Talk more on this one, or  move to bug rotation / 027 topics?
17:53:56 <armadev> and then not-entirely-consistently-used keywords indicate how far back it might or might not have been backported
17:54:04 <armadev> let's cover remaining topics
17:54:37 <armadev> then we can move to #tor-project and keep going on whichever ones we want to keep going on :)
17:55:11 <armadev> what are remaining topics? "how is the weekly bug triage position idea going" and "why no 0.2.7.7 a few weeks ago"? were there more?
17:55:39 <armadev> asn, dgoulet: how did the weekly bug triage position idea go?
17:55:48 <dgoulet> I like it personally
17:55:56 <dgoulet> and I do it here and there even if not my week :P
17:56:12 <dgoulet> teor does it also when we are asleep
17:56:13 <armadev> i think one of the goals might have been to get people more used to doing it in general
17:56:36 <armadev> i also notice that only two people signed up
17:56:45 <armadev> including into the future
17:57:11 <armadev> i will let isa and nickm ponder that one :)
17:57:14 <dgoulet> right
17:57:30 <nickm> I am indeed pondering :/
17:57:40 <dgoulet> I'm pretty sure teor didn't because he knows he could have issues guaranting some time with his contract
17:57:46 <dgoulet> didn't sign up*
17:58:45 <nickm> I'll talk to isabela and see what we come up with.
17:59:11 <nickm> on "why no 0.2.7.7" -- mainly because every time I do anything to 027, armadev asks my why.
17:59:14 <nickm> *me  why
17:59:14 <nickm> ;)
17:59:20 <nickm> and everytime I don't armadev asks me why.
17:59:21 <armadev> ha
17:59:36 <armadev> ...we should figure out how to fix that
17:59:37 <nickm> that's kind of why I want an LTS policy here, so 027 and earlier can be Not My Problem.
17:59:49 <armadev> makes sense
18:00:02 <nickm> I did release patches for 024 and forward; OpenSUSE at least applied them.
18:00:10 <nickm> okay, we're out of time.
18:00:14 <armadev> (so the answer is "actually, it was even worse than 'we had all this already backported stuff clogging up the git branch')
18:00:33 <nickm> The answer is  "You told me that all the backported stuff was clogging up the git branch"
18:00:38 <nickm> so I stopped doing 027 backports
18:01:16 <armadev> ok. is the tor browser team meeting here now? meaning we should move?
18:01:23 <nickm> I guess we should at least #endmeeting
18:01:25 <nickm> #endmeeting