09:29:50 <pastly> #startmeeting bandwidth scanning meeting 2 May 2018
09:29:50 <MeetBot> Meeting started Thu May  3 09:29:50 2018 UTC.  The chair is pastly. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:29:50 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
09:30:01 <pastly> Hello
09:30:13 <pastly> meeting pad https://pad.riseup.net/p/ioYq89yZSx1t
09:31:01 * pastly waves at teor4 and juga
09:31:08 <teor4> oh hey, I was working on the pad
09:31:54 <pastly> Don't let me pull you away
09:32:13 <juga> hi
09:33:26 <teor4> I am done
09:33:28 <teor4> on the pad
09:33:29 <pastly> When does SoP start officially?
09:34:07 <juga> pastly: something around 15th
09:34:20 <pastly> Okay thanks
09:35:25 <juga> wow, maybe i shouldn't include those much details in the pad
09:35:30 <pastly> juga: I started tagging and incrementing version numbers. I'm doing this so that it's easier to start packaging sbws?
09:35:45 <teor4> We are matching the google timeline
09:35:47 <teor4> https://developers.google.com/open-source/gsoc/timeline
09:35:56 <teor4> we are now in the "community bonding period"
09:36:31 <pastly> juga: I'd have that much detail too if I wasn't filling out the pad 10m before the meeting
09:36:35 <juga> pastly: yeah, i needed at least one version to try packaging, no need to create new ones until 1.000
09:36:54 <teor4> here is the Tor schedule
09:36:56 <teor4> https://trac.torproject.org/projects/tor/wiki/doc/gsoc
09:36:59 <pastly> juga: ok. I might keep doing it just for practice
09:37:11 <juga> hmm, i need to read what's the community bonding period
09:37:32 <pastly> thanks teor4. I just didn't see the start date on Tor's page
09:38:33 <pastly> So the transition to HTTP is going well.
09:38:52 <pastly> It accidentally also works correctly with a valid TLS cert too
09:38:57 <teor4> That's what they said in the 1990s
09:39:02 <juga> pastly: sorry, i haven't look at it yet
09:39:20 <pastly> juga: that's okay.
09:39:33 <pastly> It's all in the http* branch(es) and turning into a monster.
09:40:00 <juga> hope it'll end up simple :)
09:40:16 <pastly> As you can see here, I'm tagging ticktes I'm done with but not closing till merge. https://github.com/pastly/simple-bw-scanner/milestone/3
09:40:32 <juga> ok
09:41:14 <juga> should we go through the topics to discuss?
09:41:49 <pastly> I've avoided the consensus flavor issue by doing essentially what teor4 suggested in the trac ticket for this. https://github.com/pastly/simple-bw-scanner/blob/http_02/sbws/lib/relaylist.py#L91
09:42:00 <pastly> juga: yeah, I think we should
09:42:26 <juga> pastly: looks good
09:42:39 <juga> so, teor4, pastly what time should we then meet?
09:43:27 <pastly> I'll almost always still be driving at 1130 UTC. But 1200 would work
09:44:35 <juga> according to the pad 1200 wouldn't work
09:44:48 <teor4> That's a bit late, can we keep it to half an hour?
09:45:19 <pastly> Yeah we can try 1200-1230
09:45:38 <juga> yes, i think we could have 30min meeting
09:45:51 <juga> next topic?
09:45:57 <pastly> sure
09:46:02 <juga> teor4: Should i create a new version, include the changes or wait for more comments?
09:46:38 <juga> i can try to go line by line with all the suggestions
09:46:38 <teor4> Context? Of the bandwidth file spec?
09:46:45 <juga> yes
09:47:48 <juga> and i'd say yes to my question?
09:48:27 <teor4> We need to decide if we are doing a rewrite of the tor bandwidth file parsing code
09:48:51 <juga> the main change is the = for SP
09:49:20 <juga> ah, you meant "rewrite" with code, nor spec
09:49:39 <juga> if we change  = for SP, yes, we need a rewrite
09:49:52 <teor4> no, the main change is breaking backwards compatibility
09:50:42 <pastly> I've expressed that I think big changes in Tor is not good and we should just slip in with things that technically work and require the least amount of work. But ...
09:50:58 <pastly> I also want to stay out of it and can more easily make sbws do whatever thing Tor expects
09:51:34 <juga> teor4: so we should discuss more in the list the main change?
09:52:30 <juga> it'd be possible to implement it and still allow old format, though the code could end up more complicated
09:53:15 <teor4> we have to still allow the old format, because some directory authorities won't deploy sbws straight away
09:53:28 <teor4> I wrote 3 things that need to happen for a rewrite in my tor-dev email: https://lists.torproject.org/pipermail/tor-dev/2018-May/013154.html
09:53:47 <teor4> * The new format should have a new torrc option.
09:53:48 <teor4> * Tor should be modified to support the new format, and we should put time on the roadmap for people to work on implementing, testing, or reviewing it.
09:53:48 <teor4> * Either we should backport the new format to the latest stable release, or sbws should produce both formats.
09:54:00 <juga> oh, i didn't see that last mail
09:54:02 <juga> sorry
09:54:08 <teor4> Those 3 things are network team decisions
09:54:23 <teor4> So someone should raise it as a roadmap item for the network team
09:55:19 <teor4> If the network team wants to put a rewrite on their roadmap, they can do some of the work to make it happen
09:55:19 <juga> hmm, ic, that could change our timing quite a lot too
09:55:42 <teor4> If they don't, then we will have a minor upgrade to the format using existing code
09:57:29 <juga> so, we need to wait also the answer to that in @tor-dev?
09:59:36 <teor4> 19:55 teor4: So someone should raise it as a roadmap item for the network team
09:59:44 <teor4> at next week's network team meeting
09:59:52 <juga> ah, ok
10:00:11 <juga> one of us?
10:00:46 <juga> (sorry, don't know how this process works)
10:01:04 <juga> in any case, i can comment it in next meeting
10:01:14 <pastly> I don't think there's a formal process
10:01:22 <teor4> I will be on holidays, but I wrote a summary in https://trac.torproject.org/projects/tor/ticket/25960#comment:9
10:01:39 <teor4> the process is that we talk about roadmap changes in the weekly network team meetings
10:01:50 <juga> ok
10:01:55 <teor4> and nothing large goes in releases unless it's on the roadmap
10:02:21 <juga> ic, thanks
10:02:58 <juga> so probably during next week, i can't do too much progress on this
10:03:31 <juga> hmm, k, changing topic
10:03:38 <juga> pastly, you had log
10:03:40 <juga> log
10:03:45 <teor4> the next network team meeting is Monday
10:03:46 <juga> arg, logo :)
10:03:53 <teor4> wait
10:04:01 * pastly waits
10:04:43 <teor4> juga: you could update the spec with all the minor changes
10:05:07 <juga> teor4: ok, will do
10:05:14 <teor4> you could also write a spec with the breaking and non-breaking changes as alternatives, and delete the ones we decide not to do
10:05:23 <teor4> but only if you have time
10:05:29 <teor4> and nothing more important to do
10:05:38 <teor4> done
10:05:54 <juga> right now the most important are the tests for the current code
10:06:07 <pastly> logo: we got some person I've never heard of offer to make sbws a logo. https://github.com/pastly/simple-bw-scanner/issues/148
10:06:13 <pastly> Do we care?
10:06:19 <juga> i meant not much progress on the "implement new version"
10:06:20 <pastly> I don't think we do
10:06:28 <juga> pastly: do you know the person?
10:06:30 <teor4> juga: I agree that tests are important
10:06:45 <pastly> juga: nope
10:07:04 <teor4> You could reply and tell them about the tor ux team
10:07:24 <teor4> If sbws wanted a logo, we could ask the ux team
10:07:30 <pastly> fair enough
10:07:34 <pastly> that works
10:07:46 <teor4> But 🤷‍♀️
10:08:19 <teor4> I don't think we need one
10:08:27 <pastly> Essentially my thoughts.
10:08:34 <pastly> I'm satisfied if logo talk ends here
10:08:46 <juga> it can always come later
10:09:10 <juga> so, something else to discuss?
10:09:51 <pastly> I hope to be done with the http code in a week
10:10:23 <pastly> That's all. Not necessarily a discussion topic
10:10:49 <pastly> Well.. I'll say that it doesn't seem like the bwauths want to be able to pin some relays.
10:10:53 <pastly> So skipping that idea.
10:11:11 <teor4> good idea
10:11:19 <pastly> And client TLS certs seems ... not wanted?
10:11:26 * pastly skims quickly
10:11:31 <juga> you can still leave that option, right?
10:12:21 <pastly> If it's easy to implement or accidentally gets implemented, yes.
10:12:25 <teor4> a client TLS cert is something that people don't want until they want it
10:12:35 <teor4> in a hurry when the system is abused
10:12:43 <juga> :)
10:13:01 <pastly> I think it will be easy
10:13:03 <teor4> unlike pinning relays, which is optional and can be solved by bad relaying
10:13:14 <pastly> Way freaking easier than pinning relay(s)
10:13:21 <teor4> any misbehaving relays (or bad exiting)
10:14:01 <pastly> So that's what's different about my implementation plan from what the proposal I wrote says
10:14:16 <pastly> <done>
10:15:03 <pastly> Anything else?
10:15:15 <juga> not from my side
10:15:46 <juga> teor?
10:16:09 <teor4> juga, you said on the pad: What this mean? ^
10:16:26 <teor4> do you want to talk about the bandwidth file parsing code?
10:17:07 <juga> yeah, though i guess better out of the meeting?, too many code questions
10:17:26 <teor4> we can always do unit tests in tor
10:17:38 <teor4> we can always do bugfixes in tor
10:17:58 <teor4> features are faster after the spec is written
10:18:10 <juga> ok, understand what you mean
10:18:45 <juga> also i guess features can not be backported
10:19:19 <teor4> it is very rare
10:19:35 <juga> ok
10:19:59 <teor4> only for security
10:20:05 <teor4> I am done
10:20:22 <pastly> me too
10:20:28 <juga> me too :)
10:20:41 <teor4> thanks!
10:20:53 <pastly> Thanks for making the meeting. See you next week at 1200 UTC and hpefully lasting only 30m
10:21:01 <pastly> #endmeeting