18:04:25 <mikeperry> #startmeeting tbb-dev
18:04:25 <MeetBot> Meeting started Tue Oct 13 18:04:25 2015 UTC.  The chair is mikeperry. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:04:25 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
18:04:42 <mikeperry> Last week, I was at the OTF meeting and travelling. As a result, I didn't get a whole lot done, but I did manage to write our status report yesterday.
18:04:49 <mikeperry> This week, I need to meet a few people in Berlin (to discuss bw auth stuff and Tor Labs). I will also follow up on #6540 so that we can try to have the cert before the end of the month. I might also take a stab at updating the TBB design doc, but I am not terribly optimistic.
18:05:15 <mikeperry> in terms of focus, there are whispers that we may have a surprise Firefox release soon, but no dates or details yet
18:05:43 <mikeperry> otherwise, top priority for everyone should be the stuff under October with your name on it: https://trac.torproject.org/projects/tor/wiki/org/meetings/2015SummerDevMeeting/Roadmap/TBB
18:06:31 <mikeperry> and ideally, we'll have that in either 5.0 or 5.5alpha by the release on 11-03
18:07:30 <mikeperry> that's it for me I think
18:08:47 * GeKo appears
18:09:02 <GeKo> sorry for being late
18:09:11 <mikeperry> np. I almost forgot entirely ;)
18:09:11 <GeKo> IRL stuff :(
18:09:25 <mikeperry> too much chaos lately
18:09:56 <mcs> r.e. #6540, we may have to rearrange some things within our disk layout in order for the signing to work (see https://bugzilla.mozilla.org/show_bug.cgi?id=1046306)  So last minute is probably a bad plan ;)
18:10:19 <GeKo> exciting!
18:10:22 <mikeperry> mcs: dag. unfortunately we are caught in Apple's bureaucracy wrt getting the cert
18:10:37 <mikeperry> we started the process mid last-month
18:10:46 <mikeperry> but they are like tripple verifying we're the real tor project or something
18:10:49 <mcs> And to be clear: I have no real knowledge other than what is in the Mozilla bug report.
18:12:17 <mikeperry> ok. well, worst case we sign the 5.5a releases and keep them separate and find out which macs they actually work on
18:12:18 * GeKo still dreams of doing OS X signing on Linux machines
18:13:00 <mikeperry> if they work on 10.6-10.9.4, I'll call that a win for now
18:13:14 <mikeperry> but we can still keep them off the download page in that case
18:15:17 <mikeperry> in fact, I will send an email in reply to that thread now. continue on without me :)
18:16:28 <mcs> I can give a brief report.
18:16:35 <mcs> This past week, Kathy and I did a review plus created a mockup for #16665.
18:16:41 <mcs> And then we spent most of our time on #12967, creating patches for Tor Launcher and our gitian build scripts.
18:16:48 <mcs> We also worked on #16620.
18:16:54 <mcs> This week we hope to finish the #12967 work and make a patch for #16620 available for review.
18:17:00 <mcs> Time permitting, we will also look at #16940 and #17313.
18:17:06 <mcs> That's all for now.
18:18:24 <qwerty1> #9659 has a possible patch
18:18:44 <GeKo> yeah, it' in my review queue
18:19:11 * arthuredelstein can go
18:19:22 <mikeperry> that one may be a good idea to have mcs/brade look at, since it is deep in the networking stack (and I think mcs has a bit of experience there?)
18:19:55 <mcs> maybe not as much as you think but we could look also
18:20:15 <mikeperry> ah, I thought you worked on Netscpe server or something. isn't it all the same? ;)
18:20:26 <mcs> sure, its all the same
18:20:27 <mcs> )
18:21:30 <GeKo> sounds good to me, i just jumped at it as i think i can reproduce this behavvior on my machine(s)
18:21:34 <GeKo> *behavior
18:21:48 <GeKo> "reproduce"
18:24:41 <arthuredelstein> So here is what I did over the past week:
18:24:52 <arthuredelstein> I wrote patches for #17220 and #15564,
18:24:57 <arthuredelstein> did some investigation of #17250, #17244 and #16725,
18:25:03 <arthuredelstein> reported #17313,
18:25:12 <arthuredelstein> worked on https://bugzilla.mozilla.org/show_bug.cgi?id=1200802
18:25:19 <arthuredelstein> and started working on #17329.
18:25:39 <arthuredelstein> I think this week I will work on some more on upstreaming Mozilla patches,
18:25:43 <arthuredelstein> and try to write patches for #17329 and #17207.
18:25:53 <GeKo> thanks
18:25:54 <arthuredelstein> If I have extra time I'll work more on font fingerprinting stuff.
18:26:18 <arthuredelstein> That's all for me
18:26:29 <GeKo> arthuredelstein: do you know if there are still esr31 patches attached to some of Mozilla's tickets?
18:27:09 <arthuredelstein> GeKo: Probably. I don't think we have posted the latest tickets for many of Mozilla's tickets.
18:27:57 <arthuredelstein> Usually it doesn't seem like much progress happens unless one of use (or Dave Huseby) submits a patch for review and makes any requested changes.
18:28:31 <GeKo> i was more wondering whether updating these is a sponsorU thing
18:29:18 <GeKo> and saying, "Yeah, we updated the patches!" while there are still esr31 ones laying around sounds, well, unfortunate :)
18:29:23 <arthuredelstein> So would the idea be to post the latest ESR38 patch on each corresponding Mozilla ticket, and obsolete any old ones?
18:29:46 <arthuredelstein> Ideally we could somehow have a link that updates to the latest version of each of our patches.
18:29:53 <arthuredelstein> So that old versions don't get picked up by accident.
18:30:00 <huseby> GeKo: I think there may be some esr31 stuff laying around
18:30:10 <huseby> the spreadsheet I maintain has been around for a few years
18:30:32 <GeKo> arthuredelstein: true. let me look at the sponsoru requirements again...
18:31:51 <GeKo> "Update our patches on the Mozilla
18:31:53 <GeKo> bugtracker with each ESR release cycle"
18:32:22 <arthuredelstein> I see. OK, I'm happy to work on that.
18:32:32 <GeKo> so, I guess updating them might be smart
18:34:37 <GeKo> here is what i did:
18:34:46 <GeKo> i worked in three areas
18:35:08 <GeKo> 1) there was admin stuff to do due to the dev meeting/a new month
18:35:34 <GeKo> 2) i did some reviews/testing/triaging (#17220, #9623, #17207
18:35:36 <GeKo> )
18:36:03 <GeKo> 3) I mainly worked on #10599 which is a PITA
18:36:20 <GeKo> but it seems i've fond workaounrd anfd patches taht allow me to buld that thing at least
18:36:23 <GeKo> *build
18:36:40 <GeKo> ugh "found workaround and patches that"
18:37:17 <GeKo> i've been working on other hardenign stuff as well that we migth include into our hardened builds
18:37:32 <GeKo> and i started with #17305
18:37:48 <GeKo> input is needed there if you have some
18:38:11 <GeKo> this week i'll mainly work on the hardening bits + the related gitian changes
18:38:39 <GeKo> and hopefully we have something working next week giving us some breathing room
18:38:44 <GeKo> that's it for me
18:40:08 * boklm can go next
18:40:25 <boklm> This past week I have been looking at mozilla marionette tests, and added marionette support in our test suite for #16009. I also fixed #17126.
18:40:29 <boklm> This week I will try to run the marionette tests on Windows, and convert some mozmill tests to marionette
18:40:37 <boklm> That's it for me
18:48:40 <mikeperry> boklm: so wrt oct end, I think we probably want to be able to claim that if you run the test suite on a given linux platform, it all passes reliably
18:49:06 <mikeperry> any tests that don't do this should be flagged and/or disabled in our testing setup
18:49:26 <boklm> ok. I will check that.
18:49:58 <mikeperry> if we can do that, I think we're good on your front for now. your second year's work will be much more involved, since we want to move towards a set of tests that can be reliably run on arbitrary windows/mac/linux setups
18:50:50 <boklm> ok
18:51:51 <mikeperry> (and long-term we should plan that the set of user-runnable tests may be smaller than the ones that can run reliably on a fixed known configuration, so we should have the ability to include/exclude tests for these two situations easily, and document it in https://trac.torproject.org/projects/tor/wiki/doc/TorBrowser/Hacking#QAandTesting)
18:52:46 <boklm> yes
18:52:57 <mikeperry> https://bugzilla.mozilla.org/show_bug.cgi?id=514067 is something you should probably look at as part of this
18:54:52 <mikeperry> though heh, that bug has been weirdly pruned of useful information
18:55:09 <mikeperry> it has a lot of dependencies though :)
18:55:38 <mikeperry> I swear it used to actually have a description, at least. odd
18:56:25 <mikeperry> ok, anything else?
18:56:32 <boklm> This seems to be about unit tests that can fail depending on local configuration
18:56:38 <mikeperry> yeah
18:57:43 * GeKo spent hours debugging this stuff in the past
18:57:45 <mikeperry> in year 1 (ie: by oct 31), we just want to be able to declare "Hey, if you run this specific linux version, all our tests should be green"
18:58:23 <boklm> ok
18:58:24 <mikeperry> and we should simply disable all the cray bullshit tests that only work if the moon is full and your system has been blessed by RMS's tears or whatever
18:59:34 <mikeperry> in year 2, we'll want two configs. One for our pre-ordained holy testing setup, and one that should work for as many users as possible. even if the latter is just simple integration tests using Marionette, we're ok
19:01:08 <boklm> ok
19:01:26 <mikeperry> are we actually done at under an hour then?
19:02:21 <mikeperry> yes other mike, yes we are.
19:02:30 <mikeperry> #endmeeting *baf*