--- Day changed Sat Jan 26 2019 00:01 -!- lnostdal [~lnostdal@77.70.119.51] has quit [Ping timeout: 240 seconds] 00:39 -!- lnostdal [~lnostdal@77.70.119.51] has joined #joinmarket 02:14 -!- apeng [~apeng@unaffiliated/apengdada] has joined #joinmarket 02:18 -!- belcher [~belcher@unaffiliated/belcher] has joined #joinmarket 03:46 < waxwing> yeah nice summary thanks AlexCato 03:47 < waxwing> intriguing symptoms to say the least. 03:49 -!- GitHub154 [GitHub154@gateway/service/github.com/x-wmusweqppdmmjqxr] has joined #joinmarket 03:49 < GitHub154> [joinmarket-clientserver] AdamISZ pushed 2 new commits to master: https://git.io/fhKMb 03:49 < GitHub154> joinmarket-clientserver/master 3bfc39a AlexCato: Remove agora from default IRC configuration... 03:49 < GitHub154> joinmarket-clientserver/master 64cd0b6 AdamISZ: Merge #317: Remove agora from default IRC configuration... 03:49 -!- GitHub154 [GitHub154@gateway/service/github.com/x-wmusweqppdmmjqxr] has left #joinmarket [] 03:49 -!- GitHub136 [GitHub136@gateway/service/github.com/x-jbvmcebyqehikkxg] has joined #joinmarket 03:49 < GitHub136> [joinmarket-clientserver] AdamISZ closed pull request #317: Remove agora from default IRC configuration (master...patch-6) https://git.io/fhKnN 03:49 -!- GitHub136 [GitHub136@gateway/service/github.com/x-jbvmcebyqehikkxg] has left #joinmarket [] 03:50 -!- GitHub95 [GitHub95@gateway/service/github.com/x-nsiqmrklfnvhqofp] has joined #joinmarket 03:50 < GitHub95> [joinmarket-clientserver] AdamISZ pushed 2 new commits to master: https://git.io/fhKMA 03:50 < GitHub95> joinmarket-clientserver/master 9deb3e5 AlexCato: Do not raise exception when privmsg fails... 03:50 < GitHub95> joinmarket-clientserver/master c5fb40a AdamISZ: Merge #318: Do not raise exception when privmsg fails... 03:50 -!- GitHub95 [GitHub95@gateway/service/github.com/x-nsiqmrklfnvhqofp] has left #joinmarket [] 03:50 -!- GitHub48 [GitHub48@gateway/service/github.com/x-hxnreczktojpgmot] has joined #joinmarket 03:50 < GitHub48> [joinmarket-clientserver] AdamISZ closed pull request #318: Do not raise exception when privmsg fails (master...patch-7) https://git.io/fhK8U 03:50 -!- GitHub48 [GitHub48@gateway/service/github.com/x-hxnreczktojpgmot] has left #joinmarket [] 04:11 -!- takinbo [sid19838@gateway/web/irccloud.com/x-vyuvelisdlbcrdve] has quit [Quit: Connection closed for inactivity] 04:14 -!- apeng [~apeng@unaffiliated/apengdada] has quit [Quit: leaving] 06:57 -!- takinbo [uid19838@gateway/web/irccloud.com/x-qerinmpzwzsmzeoc] has joined #joinmarket 09:07 -!- apeng [~apeng@unaffiliated/apengdada] has joined #joinmarket 11:22 < waxwing> alexcato i think i've cracked it. all i can say right now is thank god for our regtest + miniircd test harness, no matter how crap it is. 11:23 < waxwing> the tldr is that the logic in prepare_privmsg is wrong. it uses the entry in active_channels even when the method is called with a specific mc value. after changing that, i was able to see correct behaviour after killing 1/3 message channels during the tx run. 11:26 < waxwing> the effect of the bug is something like this in broad terms: when at the start of a taker's workflow, it !orderbooks and the makers come back with their offers in privmsg, they only choose the current active_channel entry for the taker. so the taker never sees their nick on the other message channels. so nicks_seen for each message channel doesn't have all of them. then when a channel drops, the taker doesn't have any alternative channels for 11:26 < waxwing> those makers who were on that one (which one they're active on is non-deterministic, as it says in one of the comments) 11:34 < waxwing> perhaps clearer: taker !orderbook-s on all 3. so makers see taker on all 3. then they try to privmsg a response on all 3, but the logic in that function bypasses the chosen mc and uses the active_channel entry. 11:35 < waxwing> then, that is passed into signature verification with the hostid of the active_channels entry; it gets passed via daemon to client_protocol, which does bitcoin sig verification, passes back to daemon, which then calls MessageChannelCollection.on_verified_privmsg with only that hostid. 11:35 < waxwing> so the privmsg only gets sent out on that one messagechannel, with that one hostid, not all 2/3 as intended. 11:42 < waxwing> yeah confirmed i think; done multiple expts with 1/3 cut out during processing and then 1/2 also. seems to work as intended now. 12:02 -!- GitHub162 [GitHub162@gateway/service/github.com/x-cdskmnhugkojhlfr] has joined #joinmarket 12:02 < GitHub162> [joinmarket-clientserver] AdamISZ opened pull request #319: Fix bug in channel selection in prepare_privmsg (master...fix_prepare_privmsg) https://git.io/fh6fc 12:02 -!- GitHub162 [GitHub162@gateway/service/github.com/x-cdskmnhugkojhlfr] has left #joinmarket [] 12:10 < waxwing> if anyone wants to run that PR that'd be useful. i've already started running it on mainnet with all 3 servers, seems fine so far. 13:11 < waxwing> one more test: i tried doing a tx with 2 of my own wallets+bots and using -P to pick myself (as yesterday), using the PR. as before agora disconnects just after choosing the offer and starting; and it behaves as expected, correctly completing the tx on the other mchans. 13:13 < waxwing> if anyone could have a read through of #310 and leave any comments that'd be great. if not i'll read it myself at some point next week and merge d3spwn 14:30 < waxwing> finally got the http-> https redirect at joinmarket.me .. sorry about that :) weirdly i'd just commented out the relevant line in the config, and no idea why i did that. 14:50 -!- apeng [~apeng@unaffiliated/apengdada] has quit [Read error: Connection reset by peer] 16:39 < nkuttler> (http->https might leak data, if urls ever contain sensitive info) 16:39 -!- qubenix [~qubenix@66.172.11.228] has quit [Quit: quit] 16:48 < waxwing> nkuttler, interesting; you mean like for logins? 16:48 -!- qubenix [~qubenix@66.172.11.228] has joined #joinmarket 16:48 < waxwing> i'm surely being dumb but i don't get it 16:48 < waxwing> oh like if something is intended to be https and people just use the http version, i think i see what you're meaning. 16:50 < nkuttler> yup 16:51 < nkuttler> but you'd probably have to trick somebody to use the http link in the first place.. it's not a very plausible leak 16:51 < nkuttler> because if you need to trick them, you need to create the link 16:53 < nkuttler> although, some people could just type an address into the location bar and that defaults to http 16:53 < nkuttler> by redirecting http->https that use case becomes possible. dunno :) 16:53 < waxwing> yes i don't have any login thing there, but it's a good point to bear in mind. what's the standard solution? 16:54 < nkuttler> i'm not aware of a good solution for this, other than disabling http altogether 16:55 < nkuttler> or, perhaps, simply redirecting to / instead of forwarding.. actually... i could do that at work, where we're paranoid 17:07 -!- AgoraRelay [~jmrelayfn@p5DE4AD84.dip0.t-ipconnect.de] has quit [Ping timeout: 250 seconds] 17:19 -!- AgoraRelay [~jmrelayfn@p5DE4A468.dip0.t-ipconnect.de] has joined #joinmarket 21:28 < az> Hello, I've some questions about JoinMarket. Can any one join? is it profitable? What are privacy risks? 21:28 -!- Burrito [~Burrito@unaffiliated/burrito] has joined #joinmarket 21:28 -!- Burrito [~Burrito@unaffiliated/burrito] has left #joinmarket ["Leaving"] 21:31 -!- arubi [~ese168@gateway/tor-sasl/ese168] has quit [Ping timeout: 256 seconds] 21:32 -!- arubi [~ese168@gateway/tor-sasl/ese168] has joined #joinmarket 22:04 -!- v_unimportant_pe [~user@45.74.60.135] has quit [Ping timeout: 240 seconds] 22:09 -!- v_unimportant_pe [~user@45.74.60.133] has joined #joinmarket 22:20 -!- apeng [~apeng@unaffiliated/apengdada] has joined #joinmarket 22:37 -!- grubles [~grubles@unaffiliated/grubles] has quit [Remote host closed the connection]