--- Day changed Sat Aug 17 2019 00:28 -!- undeath [~undeath@hashcat/team/undeath] has joined #joinmarket 01:00 -!- Breeze555 [5c28f988@92.40.249.136.threembb.co.uk] has joined #joinmarket 01:04 < Breeze555> Hello? 01:04 < Breeze555> Anyone there? 01:12 -!- Breeze555 [5c28f988@92.40.249.136.threembb.co.uk] has quit [Remote host closed the connection] 02:20 < CgRelayBot> [cgan/AlexCato] still around, Breeze555? Just ask your question, someone will get to it 02:21 < belcher> he's gone 03:32 -!- dopplerg- [~dop@98.142.220.42] has quit [Ping timeout: 244 seconds] 03:41 -!- dopplergange [~dop@98.142.220.42] has joined #joinmarket 05:29 -!- lnostdal [~lnostdal@77.70.119.51] has joined #joinmarket 05:32 -!- kristapsk [~KK@gateway/tor-sasl/kristapsk] has quit [Remote host closed the connection] 09:14 -!- puddinpop [~puddinpop@unaffiliated/puddinpop] has quit [Ping timeout: 250 seconds] 09:19 -!- puddinpop [~puddinpop@unaffiliated/puddinpop] has joined #joinmarket 12:20 -!- belcher [~belcher@unaffiliated/belcher] has quit [Read error: Connection reset by peer] 12:20 -!- reallll [~belcher@unaffiliated/belcher] has joined #joinmarket 12:32 < waxwing> undeath, ping 12:44 < waxwing> arubi, ping, any idea on that one? https://travis-ci.org/JoinMarket-Org/joinmarket-clientserver/jobs/573215714#L2108 12:49 < waxwing> seems i have pytest 4.1 locally and 5.1 gets downloaded in travis 12:50 < undeath> waxwing: what's up? 12:51 < waxwing> dammit they deprecated it, how annoying: https://docs.pytest.org/en/latest/deprecations.html#raises-message-deprecated 12:52 < undeath> we can just remove the messages, maybe add them as a comment instead if the message seems useful 12:53 < waxwing> right that seems to be the point made in that documentation 12:53 < waxwing> shall we just follow the pattern in their example then undeath ? 12:54 < undeath> yeah, can't see why not 12:54 < waxwing> ok will do 13:15 -!- fishguts [~fishguts@121-200-12-43.79c80c.syd.nbn.aussiebb.net] has quit [Ping timeout: 264 seconds] 13:47 -!- fiatjaf [~fiatjaf@162.243.220.95] has joined #joinmarket 13:51 < waxwing> AlexCato you around? 13:58 -!- reallll is now known as belcher 14:10 < CgRelayBot> [cgan/AlexCato] yeah waxwing, what's up? 14:12 < waxwing> was just checking, i'm trying to read through #367 again, i may have a question or two over the next few hours 14:13 < waxwing> oh i guess it's late 14:13 < waxwing> sorry timezones. whatever. i'll be continuing tomorrow anyway. 14:13 < CgRelayBot> [cgan/AlexCato] sure, i'll be mostly not at home tomorrow though. But feel free to ping me on irc or reply directly in the issue, i'll see it 14:14 -!- jonatack [~jon@2a01:e35:8aba:8220:6627:dad:d967:649d] has joined #joinmarket 14:22 < waxwing> i think the sendpayment check is fine now i look at it carefully. it's fairly weak but it will cover the case where someone uses a stupid small coinjoin amount. 14:23 < waxwing> the logic we have of 'if it's -1 then take the max of (it and another calc)' is a bit weird but it's not wrong and it was there before anyway, so shrug. 14:25 < waxwing> yeah i guess it's fine. i'm going to test it in tumbler now, the most likely problem to exist here is that there's some edge case in that tumbler code that causes a crash. 14:25 < waxwing> although i can't see any. 14:25 < CgRelayBot> [cgan/AlexCato] i can always make the code more verbose and more easily understandable... performance isnt an issue for JM for these checks. So if you prefer that, I'm happy to update 14:25 < waxwing> we need to rationalise fee code quite badly, but that can be done later. 14:25 < waxwing> no the code isn't complex at all really. it's more about understanding all these different fee variables ... 14:27 < CgRelayBot> [cgan/AlexCato] yeah, i actually also agree with felix34 earlier (or was it yesterday?) that its confusing that the command line parameters for txfee do something different as well; i forgot that tbh. Sounds like something i should refactor once i have some time on my hands 14:28 < CgRelayBot> [cgan/AlexCato] just one sanity check for absolute and one for relative tx fee thresholds... shouldnt be hard to do 14:28 < waxwing> yes that was a very correct comment he made, i also agree that should be changed to be accessible on command line. 14:28 < waxwing> well yes but don't forget that what always made this complex is the necessity to estimate the fee before knowing the inputs from the makers. 14:28 < waxwing> and that whole difficulty of sweeps which are of necessity a loop solver type thing 14:29 < waxwing> so there's like target blocks and network fee estimates, fee contributions from makers, tx fee in-advance estimates without knowing inputs, sweep estimates and then two different sanity checks to avoid spending too much on fees. 14:29 < waxwing> and that isn't even a full list of it I think. 14:31 < CgRelayBot> [cgan/AlexCato] well, for tx fee estimates (used as sanity checks), it doesnt really matter if it's off by a bit; just big mistakes shall be caught by this. And these checks must happen before the tumbler starts, cant stop at random times within a tumble... so there will always be uncertainty at this point in time 14:31 < CgRelayBot> [cgan/AlexCato] which i think is unavoidable for this case 14:37 < waxwing> yes 14:37 < waxwing> wasn't talking about this specific code, talking about the context of fragmented and extremely complex rules around fees. 14:37 < waxwing> too much cognitive load for anyone trying to read the code, and good breeding ground for mistakes. etc. you know the drill. 14:42 < waxwing> alexcato : wallet.get_balance_by_mixdepth will crash with zero division if the user incorrectly chooses an empty lowest mixdepth. i'm not sure this is an issue with your code though lol (we should catch that somewhere else i think). 14:46 < waxwing> oh also, are you assumign that the entire tumbler run is funded from the starting mixdepth? ah i see you are. that's not a good assumption. 14:55 -!- undeath [~undeath@hashcat/team/undeath] has quit [Quit: WeeChat 2.4] 15:15 -!- orangesun [~aerolugge@gateway/tor-sasl/orangesun] has joined #joinmarket 16:27 -!- orangesun [~aerolugge@gateway/tor-sasl/orangesun] has quit [Quit: orangesun] 17:23 -!- AgoraRelay [~jmrelayfn@p54866090.dip0.t-ipconnect.de] has quit [Ping timeout: 246 seconds] 17:24 -!- CgRelayBot [~CgRelayBo@p54866090.dip0.t-ipconnect.de] has quit [Ping timeout: 268 seconds] 17:37 -!- CgRelayBot [~CgRelayBo@p5486675B.dip0.t-ipconnect.de] has joined #joinmarket 17:37 -!- AgoraRelay [~jmrelayfn@p5486675B.dip0.t-ipconnect.de] has joined #joinmarket 23:04 < CgRelayBot> [cgan/AlexCato] you're right, waxwing, thats indeed something I need to fix (tumbler run fee estimation only based on mixdepth 0). Hope to have time tonight to fix that 23:19 < arubi> waxwing, oof about the pytest deprecation thing. hopefully it's not a lot of work