17:59:42 <phw> #startmeeting anti-censorship team meeting
17:59:42 <MeetBot> Meeting started Thu Nov 21 17:59:42 2019 UTC.  The chair is phw. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:59:42 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:59:45 <phw> hi everyone!
17:59:52 <dcf1> hey
17:59:57 <phw> here's our meeting pad: https://pad.riseup.net/p/tor-censorship-2019-keep
18:00:40 <phw> i deleted last week's nextcloud announcement
18:01:01 <phw> shall we jump right in to the first discussion item? it's gettor's survival guide
18:01:12 <phw> but now i wonder if this was also from last week
18:02:28 <phw> the question about what's the canonical gettor repository is a good one, though. did you find out, cohosh?
18:04:31 <phw> cohosh: ^
18:05:59 * phw takes a look at the censoredplanet/russia link in the meanwhile
18:06:23 <dcf1> There's a paper attached, but I haven't read the paper yet, only the web summary.
18:06:58 <phw> oh, i didn't realise that the paper was out already
18:07:57 <gaba> hi! o/
18:08:08 <phw> hi gaba!
18:08:13 <dcf1> https://censoredplanet.org/assets/russia.pdf
18:09:17 <dcf1> One of the main observations is that censorship is less centrally managed in Russia than in China, but nevertheless effective.
18:09:44 <dcf1> Another is that Russia has the unique(?) characteristic of publishing its list of sites that ISPs are obligated to block.
18:09:57 <phw> oh, that's interesting
18:11:37 <phw> on a somewhat related note: i was upset to learn that elsevier somehow got two austrian isps to block domain names for sci-hub
18:11:46 <phw> ( https://www.derstandard.at/story/2000111305516/netzsperre-a1-und-t-mobile-blockieren-pirate-bay-der-wissenschaft -- in german, though )
18:11:59 <dcf1> Good old Wissenschaft
18:12:26 <phw> putting the 'schaf' in 'wissenschaft'
18:12:47 <phw> (the two isps, that is)
18:14:41 <cohosh> hey sorry!
18:14:46 <cohosh> i lost track of time >.<
18:15:19 <cohosh> yeah i have some questions for hiro about gettor
18:15:24 <cohosh> but if they are not here now i can ask later
18:16:16 <phw> ok, let's discuss #29207?
18:16:27 <phw> and the drop in proxies
18:16:31 <hiro> I am here
18:17:09 <phw> hi hiro!
18:17:21 <cohosh> about the snowflake proxies, it's possible that most were cupcake proxies
18:18:38 <dcf1> cohosh: yeah, that's the only explanation I can think of. WebExtension proxies should have been updated through the browser, and hosted proxies (if there are any) would have restarted themselves within a day.
18:18:58 <dcf1> I'm wondering if there's something we did not account for, though, because 90% Cupcake proxies seems high to me.
18:19:11 <cohosh> dcf1: are you getting this data from collector?
18:19:14 <cohosh> or from the debug page?
18:19:18 <dcf1> debug page
18:19:26 <cohosh> because the browser proxies are also polling only every 300s
18:19:33 <cohosh> so in reality we must have much more than that
18:19:43 <dcf1> Oh hmm, tat coudl really explain it.
18:19:55 <cohosh> there may be some conflation between #29207 and the update that extended the poll period
18:20:08 <cohosh> i just looked though and collector seems to not have the last few days of snowflake data
18:20:11 <cohosh> perhaps there is a bug there
18:20:12 <phw> wouldn't we expect most of our proxies to be firefox? after the surge of roger's defcon talk?
18:20:17 <dcf1> Okay, I'll look at the broker code to check, but I think this is resolved.
18:20:35 <dcf1> Firefox has 2,127 users and Chrome 1,921 according to
18:20:41 <dcf1> https://addons.mozilla.org/en-US/firefox/addon/torproject-snowflake/
18:20:43 <dcf1> https://chrome.google.com/webstore/detail/snowflake/mafpmfcccpbjnhfhjnllmmalhifmlcie
18:20:51 <cohosh> wow nice
18:20:58 <phw> oh, that's very helpful
18:21:41 <dcf1> Cupcake is 2,102 https://chrome.google.com/webstore/detail/cupcake/dajjbehmbnbppjkcnpdkaniapgdppdnc
18:21:45 <phw> i assume "users" means "currently using" and not "installed at some point, and possible uninstalled since"
18:22:09 <cohosh> i actually expected the cupcake proxies to stop polling though due to the deadlock bug we patched in the summer
18:22:25 <dcf1> I assume so too, otherwise what's the point of all the telemetry Mozilla and Google cram into their browsers.
18:22:28 <cohosh> so i would expect, looking at collector, for the numbers not to have changed much after #29207
18:22:53 <dcf1> Right, I report that number because that's what made me think that 90% Cupcake was implausible. But I think your polling interval change explanation is right.
18:24:35 <cohosh> hiro: hey! i noticed that the main gettor repo at https://dip.torproject.org/torproject/anti-censorship/gettor-project/gettor is behind your personal clone
18:24:50 <hiro> yes I haven't merged
18:25:05 <hiro> that's why I need to act on your reviews
18:25:14 <cohosh> i think it's farther behind than that
18:25:21 <hiro> uhm...
18:25:24 <cohosh> the deployed version is from october
18:25:31 <cohosh> and the main repo has the last commit in september
18:25:36 <phw> i'm lost here: how does the polling interval affect the proxy count on the debug page?
18:25:52 <cohosh> phw: the debug page shows the currently polling proxies
18:25:57 <hiro> I do not deploy my personal project
18:26:02 <cohosh> oh okay
18:26:06 <cohosh> maybe i'm mistaken
18:26:14 <phw> cohosh: ah, gotcha
18:26:17 <hiro> I deploy the stuff that is on git.tp.o
18:26:26 <hiro> or gitweb for the web interface
18:26:33 <cohosh> oh then maybe dip is behind git.tp.o
18:26:44 <hiro> it should be updated... but I can check that
18:27:03 <hiro> I mean automagically updated with the git hooks
18:29:00 <cohosh> hiro: oh nice, yeah i'm seeing the last commit to be august instead of oct 18 for the git.tp.o repo
18:29:24 <hiro> I have done a deploy in october
18:29:29 <hiro> let me see
18:29:29 <cohosh> i wanted to check before making a merge request to dip
18:29:56 <hiro> yeah 18-10 on gitweb
18:29:59 <cohosh> oops i should have phrased that better: the last commit on dip is august but git.tp.o is oct
18:30:08 <hiro> ah I see
18:30:23 <hiro> maybe something is not being synced I will check
18:30:28 <cohosh> ok thanks!
18:32:21 <hiro> I think it is because things moved within dip... now that I think about it
18:32:39 <hiro> gitlab should give a warning and redirect but maybe that broke the git hook
18:32:57 <cohosh> oh okay
18:33:37 <cohosh> that's all i had for discussion, sorry for jumping in late and then derailing
18:34:12 <phw> okay, let's do reviews
18:35:11 <phw> we have #32480, #31157, #30579 and an authentication token for cohosh
18:35:35 <dcf1> I'm swamped this week and cannot afford to do much review.
18:36:07 <phw> i think i can take all of these
18:36:17 <phw> except the authentication token :)
18:36:24 <cohosh> phw: thanks!
18:37:07 <phw> anything else left to talk about? if not, i'll end the meeting in a minute
18:38:16 <phw> #endmeeting