17:00:04 <nickm> #startmeeting weekly network team meeting, 3 July 2017
17:00:04 <MeetBot> Meeting started Mon Jul  3 17:00:04 2017 UTC.  The chair is nickm. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:04 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:00:07 <nickm> hello hello!
17:00:10 <ahf> hello!
17:00:15 <nickm> Yawning, isabela, komlo, isis, dgoulet, asn, Sebastian, teor, mikeperry, pastly, haxxpop, armadev, ahf, catalyst
17:00:17 <isabela> oi
17:00:19 <nickm> hihi
17:00:28 <haxxpop> hello
17:00:35 <dgoulet> hi
17:01:21 <nickm> pad is here: https://pad.riseup.net/p/JJLypSkywevA
17:01:32 <nickm> tomorrow is a us holiday, so probably we're going to be missing a few folks
17:02:06 <ahf> ok
17:02:11 <pastly> I left a brief note in the pad. I took off today
17:02:15 <Yawning> hi
17:02:18 <komlo> hello!
17:02:33 <nickm> pastly: enjoy the time off!
17:02:37 <nickm> hi everybody!
17:02:52 <nickm> so on a meta level:
17:03:05 <nickm> my highest priorities right now are to wrap up 0.3.1, to ramp up 0.3.2, and to review code
17:03:34 <nickm> We put out some security releases last week, which ate some time
17:03:40 <nickm> review-group-19 is still a thing; please review stuff
17:04:11 <nickm> If tor pays you and you are an employee: please remember that timesheets and expenses for june and earlier are needed *today*, or the accounting team will be sad
17:04:27 <nickm> and those are my meta-notes
17:04:50 <nickm> let's take a minute to write updates (if you haven't done so already) and to go through other people's updates and ask general questions
17:05:05 <nickm> I think it would be good if more folks besides me asked questions and highlighted interesting stuff
17:05:45 <ahf> it's OK with harvest as long as everything is marked 'pending approval', right?
17:05:47 <nickm> dgoulet, asn: One thing I wanted to ask is, how are things looking for July?
17:05:50 <ahf> nothing more for us to do
17:05:53 <nickm> ahf: I believe so
17:05:56 <ahf> good
17:06:04 <nickm> ahf: I'm trying to get answers from accounting there myself :)
17:06:25 <nickm> dgoulet, asn: like, would a week or two of my programming time on anything be helpful?
17:06:36 <dgoulet> nickm: I put a note on my update about that... basically end of July it is our soft deadline to have a branch with a working prop224 (client/service). Doable but it's lots of work still
17:06:42 <ahf> nickm: aye, i had a lot of questions for them last week as well :-S
17:06:42 <isabela> ahf nickm - yes should be fine
17:06:47 <ahf> great
17:06:55 <dgoulet> nickm: your review on the groundwork ticket and upstreaming stuff would be _very_ helpful for us
17:06:55 <nickm> ahf: they should make a FAQ
17:07:07 <ahf> nickm: :-)
17:07:08 <isabela> nickm: +1
17:07:13 <nickm> ok, i'll try to do that today if i don't get eaten by the holiday
17:07:17 <dgoulet> nickm: most of the groundwork ticekt (#21888) are in merge_ready and the e2e ticket is in need review for you
17:07:31 <nickm> which is the e2e one?
17:07:38 <dgoulet> nickm: #21859 ( I think)
17:07:40 <dgoulet> boom
17:07:51 <nickm> ack
17:08:12 <dgoulet> nickm: me and asn are currently in the review process of some part of the service implementation so it's going well... while asn reviews now, I'm finishing client side
17:08:15 <nickm> dgoulet: are you budgeting august for review, merge, testing, bugfixing, etc?
17:08:27 <dgoulet> nickm: most likely yes
17:08:29 <nickm> great
17:09:21 <nickm> ahf, isabela: does isabela now have the info she needs for sponsor4 reporting for june?
17:09:48 <ahf> if not, i will happily help out
17:09:54 <isabela> i believe so
17:10:02 <isabela> i will share a pad with you both too
17:10:02 <ahf> isabela: i have some spreadsheet URL's for you that might be worth including
17:10:19 <isabela> ahf: cool, do you want to email me those?
17:10:21 <nickm> great. basically, we found some bugs affecting only the test network, and one important bug affecting other tors, and fixed them, and now stuff looks better :)
17:10:35 <isabela> :)
17:10:46 <ahf> yes! we found a bug that affected us in a negative manner and now we have fixed it. so we have data from when the bug was found and how it looks afterwards
17:10:47 <nickm> if we send the spreadsheet URLs, we should explain the cases where the test network makes the numbers look worse.
17:10:56 <ahf> aye, agreed
17:11:22 <ahf> isabela: will send them to you here after the meeting
17:11:27 <isabela> ok
17:11:59 <isabela> was this bug related with one of the proposals you implemented?
17:12:05 <isabela> 140 or 278
17:12:13 <nickm> catalyst: did you see the backlog from the torlauncher UI meeting on Friday?
17:12:15 <ahf> hmmm
17:12:29 <nickm> isabela: the bug was actually related to the intersection of those proposals
17:12:39 <isabela> ok
17:12:44 <nickm> IMO
17:12:44 <ahf> it was a little bit related to all of them i think. we didn't send the "not modified" header so relays would keep asking for data they already had and receive the document
17:12:53 <ahf> thus leading to more bandwidth usage (where we want it to be less)
17:13:04 <isabela> aha, gotcha
17:13:19 <ahf> isabela: bug #22702 is where we tracked it
17:13:30 <nickm> isabela: since you're offline for the rest of the week, what should you and I do today to get your draft time stuff circulated to the team?
17:13:53 <nickm> isabela: it took ahf and me a while to figure it out though.
17:13:57 <Yawning> omg yes, i am so smrt
17:14:36 <Yawning> (I will assume that trac failing to load ever other link I click is trac being trac, and not the horrific things I just did to firefox)
17:14:38 <isabela> nickm: do you have any questions on how i set that up? maybe we can sync after this meeting? or after your lunch if you havent eaten yet
17:14:49 <nickm> after this meeting would work
17:14:57 <isabela> k
17:17:34 <komlo> haxxpop: re fuzzing framework- as you understand it more, if you can improve the documentation to make it easier for those after you, that would be excellent :)
17:17:40 <nickm> +1
17:19:11 <nickm> haxxpop: btw, I did some fuzzing ove rthe weekend; I'll write up my notes as an example
17:19:18 <nickm> (on tor_inet_pton)
17:19:56 <komlo> nickm: that would be great
17:20:29 <nickm> komlo: re mocking -- I think that usual OO thing is to define your objects so that they don't call 'down' to a lower layer, but rather so they take the object from the lower layer as an input when you're constructing them
17:20:33 <nickm> not sure if that applies here
17:20:38 <nickm> not sure what rust folks are doing
17:20:42 <nickm> please let me know what you find out
17:20:53 <nickm> catalyst: did you see the question I had for you above?
17:21:17 <nickm> catalyst: oh wait you were _at_ that meeting
17:21:20 <nickm> ug, sorry
17:21:20 <catalyst> nickm: which backlog? i think i was there for that meeting and afterward
17:21:25 <nickm> sorry there
17:21:29 <nickm> my mind is a fog
17:21:44 <nickm> isabela: what should people expect from us today wrt the time allocation thing?
17:22:02 <nickm> (and does anybody have any questions for anyone else?)
17:22:17 * dgoulet is good
17:22:18 <isabela> i believe we will share that spreadsheet and they should review as it will be how to report times from july forward
17:22:21 <isabela> till next update
17:22:32 <nickm> "how to report times, and how to allocate time"
17:22:37 <isabela> allocate
17:22:38 <isabela> sorry
17:22:42 <isabela> i mean
17:22:49 <nickm> it's important to do X if we want to report X :)
17:23:03 <komlo> nickm: yep, will investigate more
17:23:05 <isabela> yes, but we have more flexibility on that
17:23:19 <isabela> is what i meant to explain on my email
17:23:42 <ahf> is this email out already or is it something we'll receive later today?
17:23:55 <isabela> ahf: sorry an email i sent to nick
17:23:58 <isabela> not to the list
17:24:01 <ahf> oh, ok!
17:24:08 <isabela> i wanted him to review and ask me questions before senidng to the list
17:24:11 <isabela> *sending
17:25:02 <nickm> any more questions for one another or are we on to discussion topics?
17:25:06 <ahf> cool cool
17:25:32 <catalyst> how do we get blog accounts? (i want to read the preview Wilmington post)
17:25:47 <isabela> catalyst: ask hiro
17:25:54 <haxxpop> nickm, Oh thank you
17:26:12 <catalyst> isabela: thanks
17:26:34 <haxxpop> komlo, sure :)
17:27:16 <nickm> I think I can add you to the blog too
17:28:20 <nickm> catalyst: ^
17:29:00 <catalyst> nickm: that would be great, thanks!
17:30:35 <nickm> okay, let's move on to the discussion topics
17:30:47 <nickm> teor notes that we might need a mobile and/or windows hire
17:30:53 <nickm> not sure what to say there
17:32:43 <nickm> second thing: TROVE-2017-007 is an openbsd-only issue. Not sure if we should do a full backport-to-024 for that, or what.  Any thoughts?  The fix is relatively clean and simple.
17:32:51 <nickm> the ticket is #22789
17:33:31 <dgoulet> is OpenBSD using a specific version in their port (if any) ?
17:33:48 <nickm> I don't know. The user reporting this was on the latest 0.3.1
17:34:23 <dgoulet> I mean this affects all OpenBSD relays back to 024 so I would argue that backporting is good
17:34:29 <nickm> hiro: I added catalyst as a blog user. please let me know if i did it wrong!
17:34:37 <nickm> seems plausible.
17:34:50 <ahf> hm, the user is on OpenBSD-current - could be something new
17:35:01 <Yawning> http://ports.su/net/tor
17:35:06 <nickm> i wonder if this is an openbsd regression of any kind
17:35:13 <ahf> could we add a test case for it and see what the BSD builders say? i don't think any of those are running CURRENT
17:35:28 <dgoulet> oh 030 in their port
17:35:51 <nickm> i guess I could add the testcase, see if the bsd builders fail, and then add the fix?  seems a little ugly though
17:36:47 <hiro> nickm: looks fine
17:36:52 <ahf> it was more to see if it affects older versions of openbsd or if it is something new in -CURRENT
17:36:58 <nickm> hiro: ty
17:37:35 <nickm> but looking at https://buildbot.pixelminers.net/#/ , I don't see any openbsd builders right now
17:37:44 <nickm> maybe I need to write a test program and ask people to run it.
17:38:27 <nickm> I can't wait till 1 Aug when we kill off 3 old branches :)
17:39:04 <nickm> next topic is about the regression-test stuff we've been talking about
17:39:32 <ahf> hm, weird that the openbsd people aren't there, hm
17:39:50 <ahf> last seen ~24 days ago
17:39:57 <nickm> my sense is that _not_ adding regression tests when we fix something that broke is _usually_ false economy, so I think we should just declare that our rule is "when we fix a bug, make a test to make sure that the bug does not come back"
17:40:12 <nickm> we can make exceptions if there's some time when we really need to...
17:40:29 <nickm> ...but I don't think that we should pre-plan to make exceptions until we run into a case when we do truly need to.
17:40:36 <nickm> catalyst, dgoulet:  does that seem good to both of you?
17:40:49 <catalyst> mostly agree
17:41:27 <catalyst> i think making an exception process can lead people to think about whether it's more effort to write the justification or to write the regression test :)
17:41:51 <nickm> i think it will go like it went for the first 6 months of requiring unit tests for features:
17:42:01 <nickm> there will be a lot of code reviews where we remind each others...
17:42:12 <nickm> ...and where the coder decides that it's easier to just write the test
17:42:40 <catalyst> i also like teor's framing of tracking the work we're not doing
17:43:11 <nickm> dgoulet: what do you think?
17:43:27 <komlo> i agree that having a general policy of adding tests for regressions is good, maybe exceptions can be handled similarly to how not adding unit tests is handled now?
17:43:47 <komlo> if that has been a good process
17:44:15 <dgoulet> nickm: two things, I'm all for this policy, it makes sense and important (as I stated in my email). So we can try it out now and see how it goes. I've raised my concerns here about our workload so at least on its on record :)
17:45:08 <nickm> dgoulet: acknowledged!
17:45:11 <dgoulet> nickm: second thing, we should make sure the whole team is OK with it (which seems the case)
17:45:16 <nickm> +1
17:45:48 <nickm> I think that as a general rule, we should assume that all the general rules we make are subject to later discussion and critical examination.
17:46:04 <dgoulet> for sure
17:46:34 <nickm> The point of specifying these rules in advance is to work around our future shortsightedness and hubris and haste.
17:46:53 <nickm> Not to prevent us from changing our minds for _good_ reasons.
17:47:04 <nickm> Do we have more to talk about today?
17:47:19 <catalyst> also keyword tagging makes it easier for us to find the places where we cut corners where we shouldn't have :)
17:47:22 <armadev> another area to consider is external contributions. say, if somebody comes along and submits a bugfix, does that ticket go into needs-revision and they'd better come back with unit tests too? or does a core dev step in and do the unit tests? or what
17:47:48 <armadev> s/unit/regression/
17:48:06 <catalyst> armadev: i think if it's an important enough bugfix the core devs can just write the regression test if the contributor isn't prompt about providing one when asked
17:48:50 <armadev> ok
17:49:10 <nickm> I think we need to generally have a discussion about how we help people with incoming patches get their patches merged
17:49:31 <armadev> i think it's a good habit, but also i worry about reinforcing the line between "real" developers and "the rest of the world who sometimes helps"
17:49:39 <armadev> but yes, what nickm said
17:49:40 <catalyst> we could write a document that sets high-level expectations for how contributors should interact with us
17:50:00 <nickm> yes and also, our approach would benefit from something like a button saying
17:50:03 <dgoulet> probably will be case by case
17:50:14 <nickm> "That's all the hacking I want to do here! Please clean it up if you think it's worth merging"
17:50:35 <nickm> because I think volunteers who feel that don't generally say it, and we wind up politely but impatiently waiting for each other
17:50:52 <nickm> probably a good topic for a longer conversation
17:50:56 <nickm> any more for today?
17:51:21 <catalyst> #20532 replication anyone?
17:51:56 <catalyst> i wrote a recipe with what i did to trigger it and want to make sure someone else sees it too
17:52:29 <nickm> so, I've asked for repro help on the ticket...
17:52:39 <nickm> asn and I know that code best...
17:53:00 <nickm> armadev has looked at the issue too, as have dgoulet...
17:53:15 <nickm> anybody want to step up to run the recipe?
17:54:35 <nickm> ok. I will try to do it later today if I can
17:54:46 <catalyst> nickm: thanks!
17:54:51 <nickm> maybe also there would be somebody in #tor who will try?
17:55:14 <nickm> i'll ask for help there
17:55:25 <nickm> and I think it's time to end the meeting
17:55:30 <nickm> thanks, everyone!
17:55:36 <komlo> thanks!
17:55:42 <nickm> remember that tomorrow is a US holiday, and most US people will probably not want to work then
17:55:46 <nickm> peace all!
17:55:48 <nickm> #endmeeting