--- Day changed Fri Jul 20 2018
00:30 -!- puddinpop [~puddinpop@unaffiliated/puddinpop] has quit [Ping timeout: 276 seconds]     
00:31 -!- rdymac__ [uid31665@gateway/web/irccloud.com/x-sfobaprhzehbphig] has joined #joinmarket     
00:32 -!- rdymac [uid31665@gateway/web/irccloud.com/x-mtilfrlowrdzgtki] has quit [Ping timeout: 276 seconds]     
00:32 -!- rdymac__ is now known as rdymac     
01:59 -!- belcher_ is now known as beIcher     
02:38 -!- undeath [~undeath@unaffiliated/undeath] has joined #joinmarket     
07:13 < waxwing> spent some time looking at the new PSBT thing that's been merged (BIP174). it'll be something we can add to this project, as and when, basically just a standardized serialization for partially signed transactions.
07:14 < waxwing> because we don't use multisig it seems like the workflow's pretty simple; even, unchanged; taker would be 'creator', makers would be 'signer', then taker would be combiner/finalizer, something like that.
07:14 < waxwing> not even sure it's actually helpful for us to do, except in the obvious sense of standardization. the format isn't really compact, so it doesn't have that advantage.
07:33 -!- puddinpop [~puddinpop@unaffiliated/puddinpop] has joined #joinmarket     
07:51 < waxwing> undeath, i'm probably just being dumb but re: https://github.com/JoinMarket-Org/joinmarket-clientserver/pull/166/files#diff-be3c70591463c831ee6895e8c7f780ccR219
07:51 < waxwing> oh .. sorry i got it, cancel the question i was going to ask :)
07:51 < undeath> :)     
07:51 < undeath> wouldn't dare to assume you are dumb! :)
07:52 < waxwing> to keep me up to date, is there currently a form of behaviour coming out of the PR that doesn't make sense to you?
07:52 < waxwing> (i'm reading through the diff now)
07:52 < undeath> the one thing mentioned by AlexCato
07:53 < undeath> I don't know if that still happens after the last commit, I made a small change in an area that concerns code that would cause the bug
07:53 < undeath> but I can neither see how it could have happened with the initial commit, nor how the behaviour would have changed now
07:56 < undeath> currently that code is (again) in the same state as the master branch
07:56 < undeath> https://github.com/JoinMarket-Org/joinmarket-clientserver/pull/166/commits/4446537509f0c694195c80c986856b231068fb3a#diff-be3c70591463c831ee6895e8c7f780ccR248
07:56 < undeath> this is the relevant code part
07:58 < waxwing> sorry what am i looking at there? the one line highlighted (dict((v[0]...... ?
07:58 < undeath> that, including the for … sorted() loop
07:59 < undeath> the if check was removed in the latest commit but I don't see that actually chaning the behavour
08:01 < undeath> to me the behaviour described by AlexCato seems only possible if the orderbook changed 
08:01 < undeath> *offerbook
08:05 < waxwing> ok, i'm at least up to speed with what the claimed erroneous behaviour is. i'll try to check if i can see a way that's possible.
08:05 < AgoraRelay> [agora-irc/AlexCato] busy currently, but will test the changes probably later tonight. Thanks for the quick updates, undeath!
08:05 < waxwing> (you seem to be already convinced it isn't)
08:06 < undeath> thank you, AlexCato :)
08:06 < waxwing> undeath, re: the optionparser changes, thanks for doing the "merge" of the two, i think part of why i was lazy about it is, there is little reason to completely distinguish a sendpayment and tumbler function; sendpayment already supports a multi-join schedule; they should just be folded together. but, meh, it's a lot of work.
08:07 < undeath> they in fact don't seem to have that many options in common
08:08 < waxwing> well fair enough. tumbler has loads of options.
08:08 < waxwing> alexcato, when you found that strange behaviour, did you check the txfee contribution?
08:08 < waxwing> i mean, check whether it changed the ordering of which was cheapest
08:11 < waxwing> also if it's still crashing when you pick 'n', i should address that. time for mainnet testing i guess.
08:12 < waxwing> (i guess what i said above about txfee contrib is irrelevant, otherwise i'm not sure what's going on)
08:17 -!- barfbaby [a2d82ec3@gateway/web/freenode/ip.162.216.46.195] has joined #joinmarket     
08:18 < barfbaby> Hello all, I've successfully facilitated a tumble.  Now the coins are spread out across multiple internal and external wallets.  Do I need to combine those?
08:20 < undeath> you should make sure not to combine coins in different mixdepths, else you're going to render the cjs useless
08:20 < undeath> if you have coins in lower mixdepths than you need you have to cj them again
08:21 < waxwing> if you mean you finished a tumbler run barfbaby , then why are the coins in external/internal addresses in the JM wallet? they should all have gone to the destinations outside the joinmarket wallet that you specified.
08:21 < waxwing> maybe you used "destination address"es that were still inside the joinmarket wallet, is that it?
08:22 < waxwing> alexcato, undeath yes, to confirm, i screwed up the behaviour when the user chooses "no" to "accept these offers?", i'll have to push a fix for that. sorry about that ...
08:22 < AgoraRelay> [agora-irc/AlexCato] @waxwing
08:22 < waxwing> at least it's just ugly (crash message) rather than actually having any functional impact.
08:23 < AgoraRelay> [agora-irc/AlexCato] yes, txfee was negligible in both cases
08:23 < AgoraRelay> [agora-irc/AlexCato] and the cjfee difference was huge (1000 sat vs. 300000 sat).
08:23 < waxwing> it wasn't a sweep? (just to get the data exactly right)
08:24 < AgoraRelay> [agora-irc/AlexCato] it might have been, i've tested with sweeps and non-sweeps
08:25 < AgoraRelay> [agora-irc/AlexCato] if thats important, i'll remember that for tonight's testing
08:25 < undeath> I think with sweeps that might actually happen
08:25 < undeath> it doesn't seem to uniqe maker offers
08:26 < waxwing> sweeps are damn tricky
08:26 < undeath> why do they need different handling than normal cjs?
08:26 < waxwing> i remember way back, there was a case where it would completely fail to find a set if the liquidity in the pit was low. it may still be true.
08:27 < waxwing> i wrote a long explanation of the problem in a joinmarket/ Issue, let me see if i can find it
08:27 < waxwing> https://github.com/JoinMarket-Org/joinmarket/issues/356
08:27 < undeath> thanks
08:28 < waxwing> note "it filters the orders based on minsize (not maxsize, understood: because we don't yet know the cj_amount)"
08:28 < AgoraRelay> [agora-irc/AlexCato] the maker i found the error with is one of those 7.49 btc makers which tons of overlapping offers
08:28 < AgoraRelay> [agora-irc/AlexCato] might only be an issue then anyways
08:28 < waxwing> useful data, thanks
08:29 < undeath> was that changed later on, or is total_input_value not the cj amount for sweeps?
08:29 < waxwing> total input value is not the coinjoin amount, can't be right, must pay cj and btc fees
08:29 < undeath> ah     
08:30 < waxwing> sweeps are tricky because you don't know the cj amount until everything's set
08:30 < waxwing> everything being counterparties
08:30 < undeath> I see
08:31 < waxwing> i doubt i'll achieve more than fixing that "choose n" bug above today, but this finding from alexcato is interesting.
08:38 < barfbaby> To clarify, I was facilitating someone else's tumble running a YG
08:38 < barfbaby> I had coins in two wallets at depth 0 and 1
09:09 < barfbaby> I now have coins in 2 internal wallets at depth 0 and 1 external and 1 internal at depth 1
09:27 < waxwing> 2 internal wallets at depth 0? you mean, you have coins in two addresses at depth 0?
09:27 < waxwing> and this is after running the tumbler? they should all have moved out of depth 0
09:52 < waxwing> oh sorry, you're running a yieldgenerator?
09:52 < waxwing> in that case, the coins can be anywhere and everywhere, basically. they will move between the mixdepths in a somewhat random (at least, difficult to predict) manner.
09:53 < waxwing> over time they are likely to spread out over the 5 mixdepths (0..4). when coinjoins happen, they send the output (equal sized coinjoin outputs) to the *internal* branch of the "next" mixdepth, usually.
09:54 < waxwing> (although it can be coded differently, that's the default behaviour, as i recall.
09:54 < waxwing> coins will only be in the "external" branches (of any mixdepth) when you deposit them there yourself. that's the main purpose of the external branch, for deposits.
09:55 < waxwing> barfbaby, ^
10:08 < barfbaby> So perhaps the coins in the external wallet are coins I sent there and just never moved?  The rest of the internal wallets seem normal?
10:13 < waxwing> barfbaby, that should be the case, yes.
10:14 < waxwing> over time you will see more and more coins in "internal" in different branches.
10:14 < waxwing> i mean, different mixdepths.
10:14 < waxwing> you might want to open the config file (joinmarket.cfg) and set the selection algo to "greedy". it keeps the "fragmentation" lower. actually we should make that the default.
10:15 < waxwing> sorry it's called "merge_algorithm" in the file
10:45 -!- luke-jr [~luke-jr@unaffiliated/luke-jr] has quit [Excess Flood]     
10:46 -!- luke-jr [~luke-jr@unaffiliated/luke-jr] has joined #joinmarket     
10:57 -!- undeath [~undeath@unaffiliated/undeath] has quit [Quit: WeeChat 2.1]     
11:10 -!- GitHub92 [GitHub92@gateway/service/github.com/x-hzoutympoofxsvsv] has joined #joinmarket     
11:10 < GitHub92> [joinmarket-clientserver] AdamISZ pushed 1 new commit to master: https://git.io/fNCdl
11:10 < GitHub92> joinmarket-clientserver/master f1e3476 AdamISZ: Fix bug: user rejects offers resulted in crash...
11:10 -!- GitHub92 [GitHub92@gateway/service/github.com/x-hzoutympoofxsvsv] has left #joinmarket []     
11:12 -!- grubles [~grubles@unaffiliated/grubles] has quit [Quit: brb]     
11:15 < barfbaby> Thanks for the information!
11:17 -!- grubles [~grubles@unaffiliated/grubles] has joined #joinmarket     
12:56 -!- undeath [~undeath@unaffiliated/undeath] has joined #joinmarket     
12:59 -!- undeath [~undeath@unaffiliated/undeath] has quit [Read error: Connection reset by peer]     
13:01 -!- undeath [~undeath@unaffiliated/undeath] has joined #joinmarket     
13:16 -!- barfbaby [a2d82ec3@gateway/web/freenode/ip.162.216.46.195] has quit [Ping timeout: 252 seconds]     
13:31 -!- M1 [~Michail@michail.com] has quit [Quit: ZNC - http://znc.in]     
13:40 -!- M1 [~Michail@michail.com] has joined #joinmarket     
14:11 < waxwing> alexcato just popping out, but: sweep specific: see the above references about choosing based on minsize first. if the minsize is different maybe ..?
14:17 < AgoraRelay> [agora-irc/AlexCato] yes, the minsize is different in for the maker in question. Has 5+ orders, all different (but overlapping) offer sizes, differnt minsize each time
14:17 < AgoraRelay> [agora-irc/AlexCato] but the lowest minsize offer is actually a cheaper one than the one which has been selected for the join
14:18 < AgoraRelay> [agora-irc/AlexCato] ah, no.
14:18 < AgoraRelay> [agora-irc/AlexCato] My bad. Lowst minsize offer is the selected one
14:18 < AgoraRelay> [agora-irc/AlexCato] the expensive one
14:18 < AgoraRelay> [agora-irc/AlexCato] maybe whoever wrote that bot found the bug and tried to extract higher CJ fees from it
14:23 < AgoraRelay> [agora-irc/AlexCato] btw, the fix for cancelling the sendpayment works. Crashes no more.
14:26 < AgoraRelay> [agora-irc/AlexCato] ^ waxwing
15:00 -!- viasil [~viasil@134.19.180.167] has quit [Ping timeout: 268 seconds]     
15:02 -!- viasil [~viasil@134.19.180.167] has joined #joinmarket     
15:18 < waxwing> not sure if i'd call it a bug
15:19 < waxwing> in the context of weighted_order_choose it's a probabilistic advantage, but to fix it requires addressing what i described in #356
15:19 < waxwing> in the context of random under a maximum i'd agree it's a bug
15:23 < undeath> couldn't you at least remove offers by calculating with the total input and picking the cheapest fees offer of each maker?
15:23 < undeath> sure there are still edge cases but in general that should work
15:24 < undeath> and the edge cases will depend on the amount, so it's no easy to exploit
15:34 -!- undeath [~undeath@unaffiliated/undeath] has quit [Quit: WeeChat 2.1]     
16:55 -!- StopAndDecrypt [~StopAndDe@unaffiliated/stopanddecrypt] has joined #joinmarket     
17:20 -!- AgoraRelay [~jmrelayfn@p5DE4AE6E.dip0.t-ipconnect.de] has quit [Ping timeout: 256 seconds]     
17:31 -!- AgoraRelay [~jmrelayfn@p5DE4AF85.dip0.t-ipconnect.de] has joined #joinmarket     
19:01 -!- qubenix_ [~qubenix@173.199.80.131] has joined #joinmarket     
19:02 -!- qubenix [~qubenix@173.199.80.131] has quit [Ping timeout: 240 seconds]     
20:52 -!- qubenix_ is now known as qubenix