< midnightmagic> michagogo: I thought the build log getsiped everytime it tries a new stage.
< midnightmagic> wiped even
< GitHub76> [bitcoin] esainane opened pull request #7359: Disable SSLv3 for QT < 5.5 (master...patch-1) https://github.com/bitcoin/bitcoin/pull/7359
< GitHub108> [bitcoin] achow101 opened pull request #7360: Fix osslsigncode patch url (master...doc-fix) https://github.com/bitcoin/bitcoin/pull/7360
< michagogo> midnightmagic: correct
< michagogo> Wait, no
< michagogo> Build.log gets wiped once per build-script
< michagogo> So if gitian were to build on 2 archs like Linux used do, each of the two would clear build.log
< michagogo> But now build.log should cover a full gbuild
< Luke-Jr> michagogo: if you have time, would you be interested in getting the linux build to do ARM also?
< michagogo> I kinda doubt I "have time", but maybe -- what's involved?
< michagogo> And I'm still really confused how a process that's >ing to a file can just cut off/mangle the first major chunk of the file.
< midnightmagic> I have two odroid now I can build on.
< Luke-Jr> michagogo: I have no idea what's involved.
< Luke-Jr> midnightmagic: it needs to be cross-compiled, for gitian
< midnightmagic> *build for :)
< fanquake> michagogo My osx build also died last night, spun it up again now so I'll see it happens again.
< michagogo> Well, does a normal cross-build with depends work with just installing the compiler and changing the HOST=?
< michagogo> If so, gitian may be as easy as adding the arm package and adding the right triplet to the HOSTS car
< michagogo> Var
< michagogo> fanquake: you using lxc?
< Luke-Jr> michagogo: I believe so
< fanquake> michagogo: Yes
< midnightmagic> lxc on OSX? ohhhh, OSX build *under* LXC on Linux.
< michagogo> Luke-Jr: do you know the package name(s) and host triplet for arm Linux?
< michagogo> I can try it
< Luke-Jr> arm-linux-gnueabihf I guess
< Luke-Jr> and pkg gcc-arm-linux-gnueabihf ?
< Luke-Jr> not sure really
< michagogo> Okay, I guess I'll give it a try. I guess we also need pkg-config-arm-linux-gnueabihf?
< michagogo> And binutils-arm-...
< michagogo> Luke-Jr: feel free to watch as I do this
< michagogo> ssh ro-thYmcEyGTTIm0RPJm3h2bJmCp@am2.tmate.io to see me editing and running the descriptor, and ssh ro-me6PDf23hOTHOloQFCzphFCPj@fk2.tmate.io to see a tailf of the install/build logs
< michagogo> Okay, also need g++-4.8-multilib-arm-linux-gnueabihf
< michagogo> and gcc-…
< michagogo> Take 2.
< michagogo> Luke-Jr: running again
< cfields> BlueMatt: ouch, heh
< michagogo> cfields: wb
< michagogo> So, it failed again...
< michagogo> I don't get how a process >ing to a file can wipe out the beginning like that after it's already been written
< michagogo> cfields: and fanquake was also having it fail. Is anyone who has successfully built OS X using LXC?
< cfields> michagogo: i gave up on lxc ages ago
< michagogo> "Gave up"?
< cfields> michagogo: lxc has always been problematic with gitian, i go with what's stable
< michagogo> For yourself? Or for making it work at all?
< cfields> nah, just for myself
< michagogo> Well, something seems broken
< michagogo> And at least right now KVM isn't an option for me, so…
< michagogo> Until we can figure out a way to run gitian on Windows, I can only use LXC, and I imagine other people are in the same situation
< michagogo> Luke-Jr: looks like it fails in the libX11 depends build. https://www.irccloud.com/pastebin/6ct2I6Yh/
< Luke-Jr> Travis does ARM though :/
< michagogo> there's probably a way around this
< cfields> travis skips qt because X is a nightmare for arm
< cfields> NO_QT=1
< Luke-Jr> :|
< michagogo> env: HOST=arm-linux-gnueabihf PACKAGES="g++-arm-linux-gnueabihf" DEP_OPTS="NO_QT=1" GOAL="install" BITCOIN_CONFIG="--enable-glibc-back-compat --enable-reduce-exports"
< Luke-Jr> I suppose bitcoind-only builds for ARM would be an improvement still
< michagogo> I'll try to add NO_QT=1
< cfields> yea. any graphical arm build would be chipset/gpu/environment specific. just skip qt.
< cfields> for that matter, arm binaries are very much the same way. i'm not sure how much value there is in aiming for a generic one
< michagogo> Okay, take 3. Tmate sessions still live.
< Luke-Jr> cfields: it'd be nice to get one that works on RPi and Android phones
< cfields> for ex, that toolchain will limit itself to armv7 hf at a minimum, iirc. so that rules out rpi
< Luke-Jr> even the new rpi?
< cfields> Luke-Jr: exactly my point. arm != generic
< Luke-Jr> cfields: how does Android do it?
< cfields> Luke-Jr: ever use a tegra2 on android?
< Luke-Jr> no
< cfields> (the answer is no, because they didn't ship with neon, so everything broke at runtime)
< cfields> Luke-Jr: android phones are a better target, but you'll still need to set a base-line.
< cfields> cpu-intensive apps use generic builds and switch to tuned asm at runtime, based on cpu
< cfields> ffmpeg is a good example
< michagogo> cfields: did you get a chance to look for anything obvious in the log from the OS X build?
< cfields> michagogo: no, sorry. i can try to look tomorrow if it's still acting up
< michagogo> cfields: did you get a chance to see if there's anything obvious in the log from the Mac build? And is there a way to not let the early part of the log get wiped out?
< michagogo> (How is that even happening?)
< michagogo> Sorry, lag
< michagogo> Looks like using `script` should give a raw log including control characters
< michagogo> Hrm, even with no_qt=1 appended it's trying and failing to build libX11
< michagogo> Luke-Jr cfields: ^
< Luke-Jr> uppercase?
< michagogo> Yes
< michagogo> But I must have been mistaken, I was sure it failed
< michagogo> But just now it was running
< michagogo> Maybe it was frozen or something
< michagogo> Deps are built, looks like it's building okay
< michagogo> Luke-Jr: build finished fine
< michagogo> (Except that it failed at the very last line, renaming the output, but that's trivial)
< michagogo> cfields: okay, I g2g now, but I started the build again and this time I'm capturing the tailf of the build log with `script`, which should give us the whole log (I hope).
< fanquake> cfields: What's your gitian setup at the moment? I've been meaning to update mine for a while, probably worth upgrading now that I'm having an issue.
< gmaxwell> sipa: phantomcircuit: I want to give you some congrats on the a privacy improvement direct hit.
< gmaxwell> Note "compliance tracking" here https://github.com/bitcoinclassic/bitcoinclassic/pull/15 and the dishonest suggestion that restoring the trickle behavior helps '0 conf' at https://github.com/bitcoinclassic/bitcoinclassic/pull/16
< gmaxwell> kinda interesting to see 'Classic' also making privacy negative changes like XT.
< paveljanik> gmaxwell, why do you think they are going to merge this in?
< paveljanik> if gmaxwell pakes a PR to Classic uping the limit to 8MB fix, will gmaxwell' say here that they are going to merge it?
< gmaxwell> read the thread.
< paveljanik> I read it ;-)
< paveljanik> I do not care at all about Classic, but you made me curious :-)
< paveljanik> but we age going OT here ;-)
< paveljanik> but anyway, I'm happy to see you "back" ;-)
< midnightmagic> it looks like he intends to commit it to me
< paveljanik> even if so, do we care?
< haakonn> just multiple everything by 10
< paveljanik> looks like the next one will be maxconnections to 80...
< paveljanik> joined today. Some troll...
< paveljanik> 8)
< gmaxwell> sorry, I didn't mean to push this channel OT in that way.
< gmaxwell> My bringing up that repo was exclusively to point out to patrick and sipa that fixing the trickling was a meaningful enough improvement in privacy that people attacking the privacy of Bitcoin Core users are trying to undermine it.
< gmaxwell> The particular venue was uninteresting except perhaps that it wasn't at core: showing that we have a reputation for competence and caring about privacy enough that they wouldn't bother trying it on us.
< paveljanik> gmaxwell, sure. Everyone in this channel knows it...
< paveljanik> The question is how to make general public aware of this (without finger pointing).
< paveljanik> Writing articles to our web pages...
< paveljanik> e.g.
< paveljanik> blogging about it.
< paveljanik> and BTW, several days ago jonasschnelli closed plain 1M->2M PR as well... So "they" are trying to do it to Core too.
< jonasschnelli> paveljanik: how is trying what (don't read the backlog)?
< jonasschnelli> Was the close of this PR to harsh? Looked really trollish.
< jonasschnelli> s/how is trying../who s trying...
< paveljanik> jonasschnelli, I do not know who. It was just to illustrate that trolls troll this and other projects.
< paveljanik> and we can't judge the quality of the project purely on the PRs.
< paveljanik> jonasschnelli, you did perfect job in closing it...
< paveljanik> the reporter was troll
< michagogo> cfields: Yeah, as I thought, script caught this. Something really bizarre is happening when building Qt... https://usercontent.irccloud-cdn.com/file/ZiTK1ujG/1453043699.JPG
< michagogo> I'm going to try booting up my sandbox VM (that has a clean snapshot) and installing the packages to see if I can try to build outside a container
< zookolaptop> michagogo: filesystem or storage device failure ?
< zookolaptop> Sorry, gotta go play Dominions 4 with my 6-year-old. BBIAB
< michagogo> zookolaptop: nah, Windows and Linux builds are working fine
< michagogo> And the OS X build has failed in the same way at the same place 4-5 times now
< zookolaptop> Hm.
< michagogo> If you scroll back a while I uploaded the build.log
< PRab> Some quick feedback on the 0.12 RC. I'm not a big fan of the font change on the Console tab. It appears to be using a smaller font than it did before and it is harder (IMO) to read.
< PRab> (Windows 7 64bit, standard DPI)
< PRab> Another minor cosmetic issue that could be improved is that by default the User Agent column gets cut off, and the Ping Time column is huge. I would give up some of the white space in Ping Time to avoid cutting off text in User Agent.
< GitHub53> [bitcoin] pra85 opened pull request #7363: Update license year range to 2016 (master...patch-1) https://github.com/bitcoin/bitcoin/pull/7363
< adnn_> Relevant for obvious reasons.
< GitHub184> [bitcoin] MarcoFalke opened pull request #7364: [qt] Windows: Make rpcconsole monospace font larger (master...Mf1601-qtWinFontSize) https://github.com/bitcoin/bitcoin/pull/7364
< michagogo> Anyone around who gitian builds?
< PRab> I do.
< michagogo> Did the 0.12.0rc1 OS X build work for you?
< michagogo> If so, what's your gitian setup?
< PRab> I never do OS X builds. I only do windows and linux.
< michagogo> Got it.
< michagogo> No sdk?
< MarcoFalke> michagogo, you need to patch the local descriptor
< michagogo> MarcoFalke: that's just the signer IIRC
< MarcoFalke> See the 0.12 branch
< michagogo> And I'm using the descriptor from the 0.12 branch anyway
< michagogo> The issue is that it's crashing and burning in a really bizarre way when trying to build Qt
< MarcoFalke> What OS and gitian builder?
< michagogo> (See scrollback)
< michagogo> I'm using LXC running on Ubuntu Trusty in a Virtualbox
< michagogo> (on Windows, but that shouldn't matter)
< michagogo> Linux and Windows built fine
< michagogo> And I think fanquake had the same failure
< MarcoFalke> The docs mention debian, iirc. But Trusty should work as well...
< michagogo> MarcoFalke: Ubuntu is gitian's first-class platform afaik
< zookolaptop> michagogo: curious what happened with that weird truncated build log last night.
< MarcoFalke> michagogo. Do you have the build script as well?
< michagogo> zookolaptop: yeah, didn't know a truncated > file was even possible
< michagogo> MarcoFalke: sure, one sec
< michagogo> Pretty sure there's nothing abnormal there
< michagogo> And the build log looks pretty normal until the screenshot I posted
< michagogo> hmm, what's this?
< michagogo> cfields: I'm getting a 403 when depends as of v0.12.0rc1 tries to pull from the bitcoincore.org fallback server
< randy-waterhouse> hammering v0.12rc1 on the mainnet, smooth like silk ... good work guys.