IRC channel logs

2015-11-29.log

back to list of logs

***trisquel_ is now known as HeisenbergsDog
<codemac>anyone here know how to get better error messages than "source expression failed to match any pattern" ? I'm having trouble figuring out what I've done wrong, and this happens a lot while developing in guile
<lfam>codemac: You should ask in #guile. Sorry I can't help more.
<codemac>it's ok, I figured it out
<codemac>I know it means that some syntax can't be expanded or something, wish guile could print what part of the s-exp it couldn't match/expand
<zekerbauw>Hi all, i ended up with 2 GuixSD installations on my machine getting disk crypto working. One was installed from substitutes, and the other from source.
<zekerbauw>However, all the derivations, patch files, guile-builder files, etc. ended up in the /gnu/store/ of the substitute installation.
<zekerbauw>I would like to keep all of these because they are important to understand the origin of the packages on my system.
<zekerbauw>Is it bad to simply copy them into the /gnu/store/ of the GuixSD where they actually were compiled in?
<zekerbauw>I know /gnu/store is normally supposed to be read-only, but in this case the data copied in only came from Guix in the first place.
<zekerbauw>I want to run gix pull, and it would maybe take less time if there are derivations already built that i can copy back in
<jroh>anyone have any luck with setting up the guile-wm package, or getting startx going w/ a .xinitrc ?
<iyzsong>jroh: I did, but guile-wm is not useable for me. startx won't work now (I have some plan for it in my TODOs..)
<jroh>kk. any idea what other WMs than ratpoison and xfce work well atm?
<iyzsong>jroh: all should work, if it doesn't show in the slim menu, you can put it in ~/.xsession. I think openbox (non tiling) and i3 (ttiling) is good start for WMs, and Enlightenment is another choice.
<apa512>i'm new to guix and getting "bzip2: Compressed file ends unexpectedly;" while attempting a system reconfigure.
<apa512>this happens on http://hydra.gnu.org/nar/fnd9k8rfz5jbvj80zjqh6l95imckqp51-linux-libre-4.2.6
<apa512>downloading the file manually and running bunzip gives the same result. so can i assume that this is an error with the guix server and nothing i can fix?
<lfam>apa512: Most likely the download has failed partway through. That's what that error message means in my experience.
<lfam>ACTION trying it
<apa512>right. i've tried this many times, so either the hosted file is broken or something is weird with my system.
<apa512>running default desktop config + icecat
<apa512>might have happened after guix pull
<lfam>Yeah, that file is corrupted. I just downloaded it and decompression fails. Is that the most recent build? Can you check the status on hydra.gnu.org
<lfam>apa512: Can you send a bug report to bug-guix@gnu.org?
<apa512>sure
<codemac>question! is there something borked with the linux-libre package in guix master? I keep getting an unpack error on gigaset.ko for linux-libre-4.2.6
<lfam>codemac: Yes. Another user just mentioned it a few minutes ago. I asked them to file a bug report. Want to race them? ;)
<lfam>BTW codemac: How goes Go?
<codemac>Still stuck with gcc injecting libgcc_s.so deps randomly based on how go execs cgo
<codemac>A coworker has been helping me learn the compiler's implementation, so I may be able to patch things
<lfam>Wild. Do you know if Nix suffered the same problems?
<codemac>See - I haven't actually confirmed that they got cgo working
<codemac>I emailed their maintainer, but he didn't have much to say
<lfam>I use some Go software on NixOS so they did something. But I don't know how they did it. Is your latest try in the guix-pkgs repo on github? I could try to compare it to the Nix packaging and report back.
<codemac>I can check for you real quick, I'd appreciate the feedback
<jroh>codemac apa512 fwiw, I think I used --fallback to get around that error earlier.. but that meant waiting for the kernel to compile
<lfam>jroh: That indirectly answers my question about --fallback earlier. I didn't understand what it did. I interpreted it as falling back to building if substitutes weren't available, but it actually means fall back to building if substituting has an error.
<lfam>Thanks!
<lfam>I can't read source code on a bus :( I have to ask questions on IRC
<codemac>lfam: rebuildng my go package to make sure I didn't mess it up first :P
<lfam>codemac: I thought it didn't work ;)
<lfam>codemac: bbiab
<codemac>ok lfam, it's up2date, set the tmp folder as your GUIX_PACKAGE_PATH
<lfam>codemac: How'd it build?
<codemac>lafm: it goes all the way except failing the RUNPATH check for 1.5.1
<codemac>for libgcc_s
<codemac>(which is where I was before)
<codemac>you can try by setting the tmp folder in the repo to your GUIX_PACKAGE_PATH
<codemac>guix build go-1.5.1
***HeisenbergsDog is now known as Guest75575
<codemac>Can I get guix-builder to use a different folder than /tmp?
<lfam>codemac: Thanks, I'll check that out tomorrow. You can start the daemon with TMPDIR set to wherever you want.
***Guest75575 is now known as HeisenbergsDog
<ploopkazoo>is there a GuixSD channel?
<lfam>ploopkazoo: This is it
<ploopkazoo>oh
<codemac>the best part about guix is knowing that no matter how disgusting the build process is for an application, if you can get it packaged, *everyone* can do the build. This always seemed silly/stressful with arch dev for me
<efraim>using debian I was taught about the horrors of compiling software myself, that with 20,000+ packages, if it wasn't in the repos you didn't need it.
<efraim>If you did really want it anyway, you should either figure out the whole .deb packaging thing, or run `sudo make checkinstall` instead of `sudo make install` to register it with dpkg/aptitude
<HeisenbergsDog>coming from fink/macosx, i like guix exactly for the same reason
<HeisenbergsDog>even if i dislike scheme
<xd1le>efraim: how does guix packaging compare to the "whole .deb packaging thing"
<xd1le>?
<efraim>xd1le: I've looked at the debian new maintainers handbook several times, and compiling the software is the easy part, but I still have no real idea on the "best practices" for the debian folder, patches, building locally, keeping all the packaging stuff in a separate git repo, etc.
<xd1le>efraim: ah i see. i don't know anything about it hence why i asked.
<civodul>Hello Guix!
<efraim>hi!
<civodul>there's a python2/setuptools issue in tk-update, see http://hydra.gnu.org/build/835722
<civodul>iyzsong, efraim: any idea what's wrong? ↑
<civodul>this is a new failure compared to master, see http://hydra.gnu.org/eval/108145?compare=master#tabs-now-fail
<civodul>and an R failure due to rounding on MIPS: http://hydra.gnu.org/build/834308/nixlog/2/tail-reload
<efraim>I can try to take a look at it
<civodul>efraim: that'd be great! :-)
<efraim>for python2-pathpy, commit 553b709ba103a61a4a934837cb4de4b4f53665ef on the 26th adds python-setuptools as a native-input for python-pathpy
<fps>what's the bot's name again?
<fps>to tell people something later?
<fps>sneek: tell lfam later: the substituter failing is maybe different from it having no substitute available?
<sneek>lfam, fps says: later: the substituter failing is maybe different from it having no substitute available?
<fps>umm
<efraim>you can also do sneek help, and it'll whisper you the commands
<efraim>i always forget
<fps>sneek: later ask lfam: the substituter failing is maybe different from it having no substitute available [re --fallback]
<sneek>Will do.
<fps>sneek: botsnack
<sneek>:)
<efraim>python programs often have an "if python-setuptools, then import setuptools, else use nativetools" or somesuch, but nativetools/whatever its called doesn't seem to be in python2.7
<efraim>I manually added python-setuptools to python-pathpy and I'm building it locally to see if it works still
<civodul>cool
<civodul>ACTION goes afk for a couple of hours
<efraim>sneek: later tell civodul the python2-pathpy failure is fixed in master, it is just missing setuptools, and when I added it to the inputs it build without issues.
<sneek>Got it.
<efraim>sneek: later tell civodul the python-statsmodels failure (http://hydra.gnu.org/build/834189) I'm not so sure about
<sneek>Okay.
<efraim>sneek: botsnack
<sneek>:)
<ennoausberlin>Hello. Is there a way to configure keyboard layout for X and terminal in config.scm?
<sneek>Welcome back ennoausberlin, you have 1 message.
<sneek>ennoausberlin, mark_weaver says: if <OUTPUT>/share/xsessions/*.desktop is present, then slim should see it and support it as a window manager option.
<HeisenbergsDog>efraim, nice bot :)
<HeisenbergsDog>who owns/runs it?
<Digit>compiling 475 files. should i be doing guix pull more often?
<civodul>Digit: no, that's normal :-)
<sneek>civodul, you have 2 messages.
<sneek>civodul, efraim says: the python2-pathpy failure is fixed in master, it is just missing setuptools, and when I added it to the inputs it build without issues.
<sneek>civodul, efraim says: the python-statsmodels failure (http://hydra.gnu.org/build/834189) I'm not so sure about
<civodul>ok!
<civodul>thanks for investigating
<efraim>`guix environment aria2 --pure less strace` should give me an environment with aria2's inputs & less and strace only?
<civodul>yes
<civodul>you might want --container instead of --pure, esp. if you're on a foreign distro
<efraim>using the liquorix kernel on debian, if I do `guix environment vim --pure --ad-hoc which less` I get vim's inputs and also less and which
<efraim>I'll have to look at --container more to see why it's giving me troubles
<civodul>efraim: what troubles do you have?
<civodul>--container works like --pure, except that it spawns a new container
<efraim>`guix environment foo --pure bar` gives me the dependencies of foo and bar in a new shell, but neither foo or bar
<efraim>`guix environment foo --pure --ad-hoc bar` gives me the dependencies of foo, and bar itself, in a new shell
<efraim>`guix environment foo --container` doesn't return anything, and doesn't give me a new shell
<efraim>the liquorix kernel specifically doesn't react to `sudo sysctl -w kernel.unprivileged_userns_clone=1`, but does let me run VMs
<civodul>does it show an error message? what's its exit code?
<civodul>normally --container prints a helpful error message if containers aren't supported
<efraim>it doesn't print anything, it just thinks for a while and then returns the $ so I can enter another command
<efraim>up-arrow allows me to run it again, so I know its not in a container
<civodul>what's its exit code?
<civodul>that is: "guix environment foo --container ; echo $?"
<efraim>1
<civodul>uh
<civodul>could you run "strace -f -o log guix environment foo --container ; echo $?" and send the output and file 'log' to bug-guix@gnu.org, as well as information about the kernel version and Guix commit?
<civodul>looks like a genuine bug
<efraim>sure
<civodul>emacs-devel strikes again
<civodul>i wonder if i'm just being unduly irritable
<codemac>probably both
<codemac>;)
<HeisenbergsDog>heh
<civodul>:-)
<b4283>hello
<b4283>i got the "waiting for "root" device to appear" message, and after some searching i realized that there's a mistake in the config.scm
<b4283>is it fixable to correct it by simply modifying config.scm ?
<efraim>what is the log attachment limit size for bug-guix? log came out to 3.3MiB
<civodul>b4283: what does "e2label /dev/your-root-partition-device" return?
<civodul>sounds like it may be lacking the label that is specified in the config
<civodul>efraim: not sure; could you pass it through "gzip --best" and attach the result?
<efraim>good idea
<efraim>136KiB
<lfam>sneek: later tell HeisenbergsDog check out this log for more info on sneek: https://gnunet.org/bot/log/guix/2015-05-25
<sneek>Welcome back lfam, you have 1 message.
<sneek>lfam, fps says: the substituter failing is maybe different from it having no substitute available [re --fallback]
<sneek>Okay.
<lfam>fps: Yeah, that seems to be the difference
<fps>so you guys have no estimates for the amount of traffic hydra delivers each month?
<cts>Why does guix-daemon version 0.9.0 not go in the background background?
<rekado>lfam: I don't think peg and leg need to be installed. They are only used at compile time IIRC.
<rekado>peg-markdown bundles the peg+leg sources.
<lfam>rekado: Wow, I totally misunderstood it. I was looking for an install target somewhere and the only one I found installs peg and leg.
<rekado>peg-markdown produces a "markdown" executable, I think.
<rekado>I only ever used it through Ruby bindings ("parkdown").
<civodul>fps: no!
<lfam>rekado: Will you look at version 3 :( of the peg-markdown patch before I send it? https://github.com/lfam/guix/commits/contrib-peg-markdown
<rekado>lfam: I haven't build it using your package definition but it looks fine to me.
<rekado>(I'd change the indentation of the "(uri" line, though)
<rekado>don't know if the leak check is useful.
<lfam>rekado: How would you change that line?
<lfam>The leak-check phase makes it a pretty heavy package, since it requires Valgrind.
<rekado>I'd pull part of the string onto the same line as "(string-append".
<rekado>instead of the leak-check stuff you could run the conformance tests.
<rekado>"./MarkdownTest.pl --script ./markdown" or something like that.
<lfam>Alright, I'll put that in there.
<civodul>"Sorry if this process seems a bit unfriendly, by the way. Reviewing is often a pain."
<civodul>ACTION finds it disconcerting
<orbea>I dont get tit, what am I missing? http://dpaste.com/3FBTVN5
<orbea>*get it
<civodul>orbea: weird, is this on a recent checkout or 'guix pull'?
<orbea>as of a few days ago
<cehteh>cant you invent some cryptocurrency? guixcoin .. every MB compiled generates a coin?
<cehteh>srsly .. guix store and its history has properties of a blockchain
<orbea>the pull finished, no guix refresh still
<civodul>cehteh: that's an idea :-)
<civodul>orbea: what does "guix --help" return?
<orbea> http://dpaste.com/1FGEMB4
<civodul>and "guile -c '(use-modules (guix scripts refresh))'"?
<orbea>that works
<orbea>its warning about a lot of unbound variables
<civodul>and "guile -L ~/.config/guix/latest -C ~/.config/guix/latest -c '(use-modules (guix scripts refresh))'"?
<civodul>could you paste the warnings?
<orbea> http://dpaste.com/0GXQ8DA
<civodul>orbea: there's a problem here, all these things should already be compiled
<civodul>are you running in a git checkout?
<orbea> \\http://dpaste.com/2W63YQ6
<orbea>i installed it from latest tarball
<civodul>binary tarball?
<orbea>source
<civodul>and you did "make install" from there, right?
<orbea>here is the slackbuild I used http://slackbuilds.org/slackbuilds/14.1/system/guix/guix.SlackBuild
<civodul>ok
<civodul>could you run the "guile -L ..." command i gave above without the double quotes?
<orbea>oops
<civodul>:-)
<orbea> http://dpaste.com/0CAX9WK
<civodul>here we are
<civodul>but this was supposed to be fixed by b68d2db
<civodul>oooh, i see
<orbea>guix is very different than Im used to with slackware, I'm a bit lost :)
<civodul>that commit does the check at compile-time, but at run time things are different
<civodul>anyway, could you install guile-json?
<orbea>from guix or the host os?
<orbea>I'd have to make a slackbuild for guile-json I think, I can do it, but only if I need to
<civodul>orbea: i've pushed a fix for that
<civodul>so you could "guix pull" and it should work this time
<civodul>(alternately, you can install guile-json with guix)
<orbea>cool, thanks
<orbea>i managed to make a guile-json slackbuild while I waited...
***fakeuser is now known as francis7