--- Day changed Sun Dec 09 2018
01:27 -!- undeath [~undeath@hashcat/team/undeath] has joined #joinmarket     
05:22 -!- felix34_ [~muh@5.2.76.19] has joined #joinmarket     
05:23 < felix34_> The correct way to upgrade to a new release is `git pull` followed by `./install.sh`, correct?
05:27 < undeath> in my experience "python setupall.py --all" works just as well instead of running install.sh
05:28 < felix34_> ok, thanks!
05:47 < undeath> oh, and ofc if you want to run the latest tagged release instead of the current master branch you need to do a "git checkout" after the "git pull"
06:13 < felix34_> what's the benefit to that? avoiding potential bugs?
06:15 < waxwing> felix34_, yes it should be slightly safer, but we don't generally add stuff to master unless it's been tested quite a bit, and/or we feel there's no risk. current master has python3 if you're interested, it works fine unless you're using Qt, in which case it doesn't (yet).
06:16 < waxwing> if you want to install for python3 with current master, prob best is just `./install.sh --develop -p python3`
06:16 < waxwing> everything else should be the same (the jmvenv will use python3 instead of 2)
06:24 < felix34_> waxwing: python3 is excellent, but since I have no specific need of it I think I will just leave that until the next upgrade, I got the upgrade already done and as they say "if it isn't broke don't fix it"
06:26 < waxwing> felix34_, yup
07:42 -!- puddinpop [~puddinpop@unaffiliated/puddinpop] has joined #joinmarket     
08:15 -!- GitHub45 [GitHub45@gateway/service/github.com/x-gfcnsgpcpjaunxem] has joined #joinmarket     
08:15 < GitHub45> [joinmarket-clientserver] AdamISZ opened pull request #248: [WIP] Convert JoinmarketQt to PySide2, qt5reactor (master...PyQt5) https://git.io/fp9W4
08:15 -!- GitHub45 [GitHub45@gateway/service/github.com/x-gfcnsgpcpjaunxem] has left #joinmarket []     
08:38 < undeath> uh, there is another error in the test_client_protocol.py
08:39 < undeath> why did this not occur for me locally? :/
08:42 < undeath> also it's only failing on py3 but that doesn't seem like a py3-specific problem
08:44 < undeath> maybe the test twisted test runner is simply a bit better on py3
08:55 < waxwing> electrum is too advanced for me sometimes: https://github.com/spesmilo/electrum/blob/d0e6b8c89dee017e4ad790d63e4f642a7e28e67c/electrum/gui/qt/password_dialog.py#L46
08:55 < undeath> lol     
08:57 -!- arubi [~ese168@unaffiliated/ese168] has joined #joinmarket     
08:59 < undeath> what's broken with the qt5reactor on pip?
09:09 < waxwing> don't remember exactly i can check later. Lightsword might.
10:36 < Lightsword> not sure why it would be broken
10:42 < waxwing> oh i recall now what the error message is - it fails to load PyQt5, whereas the fork tries PyQt5 and then PySide2
10:48 < undeath> maybe only a problem if you have pyqt5 + pyside2 installed?
10:56 < waxwing> undeath, no the pip one just tries PyQt5, only
10:57 < undeath> oh, I see
10:58 < waxwing> so i guess the other way round, i've tried so many things now i'll probably forget all the details, but if pyside2 works let's go with that, Lightsword also thinks it's a better option (and seems that's probably the case from where i sit) to have it work on Py2 and Py3.
10:58 < waxwing> testing this is .. fun ...
10:58 < undeath> :/     
10:58 < Lightsword> I think supporting both is probably not too hard lol
10:58 < undeath> gui testing is really annoying
10:58 < Lightsword> ikr     
10:59 < Lightsword> most of the GUI’s I write are webapps
10:59 < undeath> i recall there are frameworks to automate it
11:00 < waxwing> it is possible, yes. but at least here the functional delta is basically zero. there are basically like 5 or 6 things it can do, it's a pretty simple app.
11:01 < waxwing> what's hard is not so much running it (although ofc automated's better), but having to do it across different python versions, and the fact that you kinda have to do the bulk of the work on regtest
11:01 < waxwing> .. but then also sanity check on mainnet because it's not the same (that was probably a bad idea, although it seemed convenient at the time)
11:03 < waxwing> Lightsword, if you or anyone wants to write a web app that'd be cool :) mostly it's just instantiating a taker and then giving it some callbacks. at least, to just do like a single join / Taker only.
11:04 < waxwing> Lightsword, do you happen to know if binary packaging is any easier or not with PySide2? i guess not.
11:04 < Lightsword> heh, I usually don’t even like writing webapps, just they usually are easier than stuff like qt
11:04 < waxwing> i used pyinstaller before but it got quite hard with the dependencies.
11:04 < Lightsword> I’d assume it was about the same
11:04 < waxwing> esp for windows, ofc
11:05 < Lightsword> wxwidgets was the easiest I found
11:43 < waxwing> Lightsword, appears that it's possible this can throw an IndexError: https://github.com/JoinMarket-Org/joinmarket-clientserver/blob/master/jmbitcoin/jmbitcoin/secp256k1_main.py#L115
11:45 < waxwing> if divmod behaves normally i guess that shouldn't be possible, but it seems to have happened
11:46 < Lightsword> what was the remainder from divmod where you saw the error?
11:47 < waxwing> i'll try to get it (this should prob be repeatable). weird.
11:56 < waxwing> https://pastebin.com/MfSrpqhU
11:57 < waxwing> two examples there of it segfaulting. this is happening Py2 only, not Py3.
11:57 < waxwing> i tried to get a printout of the divmod but i guess it crashed before printing. it seems pretty clear something very fundamental is wrong :)
11:58 < waxwing> like i said in the PR, this is somewhat reminiscent of issues i had with threading in the first version of this; eventually i realised i had to run everything in the reactor, not with a thread for joinmarket; then it worked (using pyqt4reactor)
11:58 < arubi> "line 139703898044224, in _commandReceived"  what the heck
11:58 < waxwing> it seems to be working similarly fine in Python3 here, with qt5reactor ... but, in Py2 seems all screwed up (although things work until i do joins, i think, so hmm maybe it has to do with when it gets to the network/irc stuff)
11:59 < waxwing> arubi, lol good point
11:59 < waxwing> makes about as much sense as getting an index error after that divmod :) (i don't *think* that's mathematically possible)
11:59 < undeath> memory corruption?
12:00 < arubi> it's just under 2^47
12:00 < arubi> 46.98936560407820691930
12:00 < arubi> log2
12:00 < arubi> yea I think you're right that it's related somehow
12:01 < waxwing> undeath, well everything else is working fine, including this code on Py3. i think some combination of py2, pyside2 and qt5reactor (may well be the latter) . well and twisted.
12:01 < undeath> also the segfault after the exception; python programs don't segfault on exceptions
12:02 < waxwing> undeath, well .. pyqt is kinda a binding or something no? i've seen segfaults with pyqt before.
12:02 < undeath> yeah, there is probably some bug in the binding
12:02 < undeath> i've seen things like this happen before
12:02 < waxwing> well, that or .. i'm using something that's not threadsafe in a thread.
12:04 < waxwing> maybe qt5reactor is just doing something naughty, a bit lost tbh.
12:05 < arubi> waxwing, I'm wondering if this is running on a 32 bit machine running a PAE enabled kernel :P
12:09 < waxwing> arubi, i've no idea what that last part is, so .. no :)
12:11 < arubi> hehe ok, was thinking maybe it was some kind of pointer in a 48 bit address space, obviously I was way off :)
12:20 -!- Giszmo [~leo@190.162.241.129] has quit [Ping timeout: 272 seconds]     
12:34 -!- Giszmo [~leo@ip-123-233.219.201.nextelmovil.cl] has joined #joinmarket     
12:46 < Lightsword> fyi the future builtins module essentially makes all integers into longs(on py3 the long type was removed and py3 ints are essentially py2 longs)
12:57 < waxwing> Lightsword, interesting, thanks
12:58 < waxwing> i tried changing to clearnet in case socks.py matters, but it didn't
12:58 < waxwing> i'm wondering about threading, although there is use within Qt itself, it's only a side-thing (privkeys export). i can't see where else that might creep in, maybe the HTTP stuff in jsonrpc.py?
13:01 -!- nothingmuch [~user@213.152.162.5] has quit [Ping timeout: 240 seconds]     
13:08 -!- GitHub174 [GitHub174@gateway/service/github.com/x-mimwmfuxsanbokml] has joined #joinmarket     
13:08 < GitHub174> [joinmarket-clientserver] undeath pushed 1 new commit to master: https://git.io/fp9gV
13:08 < GitHub174> joinmarket-clientserver/master 3ae5bff undeath: fix TestMakerClientProtocol failing with py3 (2)
13:08 -!- GitHub174 [GitHub174@gateway/service/github.com/x-mimwmfuxsanbokml] has left #joinmarket []     
13:08 < undeath> this time I made sure travis works on my fork before
14:12 -!- \x [~user@unaffiliated/-x00/x-1893888] has left #joinmarket []     
14:51 -!- undeath [~undeath@hashcat/team/undeath] has quit [Quit: WeeChat 2.3]     
15:56 < waxwing> suspicious how similar this is; even though a different error message, it's a failure on exactly the same line: https://github.com/Syncplay/syncplay/issues/147#issuecomment-336956403
16:01 < waxwing> this looks very relevant: https://bugreports.qt.io/browse/PYSIDE-629?focusedCommentId=395344&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel
16:01 < waxwing> especially because of py2.7 reference
16:04 < waxwing> given the segfaults described from here, it may well be the exact same bug https://bugreports.qt.io/browse/PYSIDE-629?focusedCommentId=407267&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-407267
17:51 -!- AgoraRelay [~jmrelayfn@p54866D7B.dip0.t-ipconnect.de] has quit [Ping timeout: 268 seconds]     
18:00 -!- AgoraRelay [~jmrelayfn@p5DE4AAA8.dip0.t-ipconnect.de] has joined #joinmarket     
18:02 -!- beIcher [~user@unaffiliated/belcher] has quit [Ping timeout: 240 seconds]     
18:08 -!- beIcher [~user@unaffiliated/belcher] has joined #joinmarket     
19:40 -!- Giszmo [~leo@ip-123-233.219.201.nextelmovil.cl] has quit [Ping timeout: 246 seconds]     
19:59 -!- Giszmo [~leo@190.162.241.129] has joined #joinmarket     
20:10 -!- nothingmuch [~user@213.152.161.85] has joined #joinmarket     
21:46 -!- midnightmagic [~midnightm@unaffiliated/midnightmagic] has quit [Ping timeout: 264 seconds]     
22:17 -!- midnightmagic [~midnightm@unaffiliated/midnightmagic] has joined #joinmarket