IRC channel logs
2018-02-11.log
back to list of logs
<marusich>Apteryx, yes, the full path is "just" another store path. <marusich>Specifically, the "out" output, I think, gets used in the calculation of the output's store path. <marusich>Therefore, if you have three outputs - "out", "debug", and "doc", for example - you would get three different store paths, I think. <marusich>In the case of outputs that are not "out", I believe we also add the output name to the path. For example, the "doc" output of "gnucash" on my system lives at /gnu/store/b7121fqqsglgf5wkhay3imf9ak076k6y-gnucash-2.6.18-doc. <marusich>And that's different from the normal "out" output path, which lives at /gnu/store/3fydrrhmwj9zhmis95rd6snd4l8958ra-gnucash-2.6.18 and has a different hash. <Apteryx>can't I use (define* with #:allow-other-keys but without defining any keyworded arg? <pkill9>it's failing in the testing phase <Apteryx>how often does guix checks for grafts? <Apteryx>(how often does it invalidates its cache) <lfam>Apteryx: Grafts come with other package changes via `guix pull` <sneek>janneke, you have 1 message. <janneke>is berlin.guixsd.org's cuirass api available? <janneke>OriansJ: thanks for your `horrible' stuff ;-), how do i use it? <efraim>you can verify the signature as a regular user <Tirifto>The stated aims for the step are: “Make root’s profile available under ~/.guix-profile” and “Source etc/profile to augment PATH and other relevant environment variables:” <efraim>the assumption is that '$' is for a regular user and '#' is for root <rekado_>Tirifto: we also have an installer script that takes care of all of these things. <Tirifto>Ohh! That makes a lot of sense and it went right over my head. <Tirifto>And within the tutorial, is $HOME as used by root supposed to point to “/root” rather than “/home/username”? <Tirifto>Excellent, that clears things up. Thank you! :) <ng0>yep, server crashed. see the email I've sent this week :/ .. I'm slowly bringing it all up again. git://code.crash.cx/ng0/guix/guix.git will work <ng0>https fetch will return eventually <efraim>I got my aarch64 build box back online finally <pkill9>is it possible to roll back a `guix pull`? <janneke>pkill9: the effect you experience from a guix pull is that your ~/.config/guix/latest symlink has been moved <mbakke>pkill9: you can run `guix pull --commit=...` <janneke>pkill9: if you reset it to its previous value, or a git checkout eg, you're fine <pkill9>hmm, it would be good if guix automatically made symlinks to previous generatiosn of guix-latest, like it does with profiles <ng0>I think someone broke the qemu build. I'm doing another build to check it <catonano>in building a python package I get NameError: name 'doctest' is not defined <Apteryx>what was the package required to get the "python" executable? <efraim>I don't have a copy of it on my aarch64 box it seems <Apteryx>what does this error mean? (502). Does the archive need to be manually cleared on the substitute server? <ng0>hm.. qemu not broken. then it was something on the server <catonano>ah I messed up some imports in the tests off my module, sorry <davidl>Im trying to use my own xmonad.hs file but when I reload the config I get the error /run/current-system/profile/bin/ld: cannot find crt1.o: No such file or directory. <davidl>I have tried setting LD_LIBRARY_PATH and LIBRARY_PATH to include the path to crt1.o but that didn't help. <davidl>Im also having the recent user-homes could not be started problem. <wigust>davidl: Do you have a gcc-toolchain package installed? <davidl>wigust: I have glibc installed. The problem went away when I appended the path to GHC_PACKAGE_PATH <davidl>wigust: so not sure actually how to check whether a package is installed without tabbing a package-related command. <davidl>although now I got a new problem instead. <davidl>wigust: ok, thanks. Ill try installing that package. <Apteryx>guix environment: error: build failed: some substitutes for the outputs of derivation `/gnu/store/kcsa9msjkv2c3i07h7cc58b0m4dj610r-tiff-4.0.8.tar.xz.drv' failed (usually happens due to networking issues); try `--fallback' to build derivation from source <Apteryx>maybe some recent development on Cuirass? <ngz>I seem to encounter an error with Guix recently. Whenever I add a package to my profile, I end up with a bunch of "wrong ELF class: ELFCLASS32" errors related to gtk. <ngz>Is it something known? <davidl>wigust: now I can recompile but during login it complains about not finding libxft.so.2 and then fails. <davidl>I tried the fc-cache -f and installing font-packages <pkill9>ngz: are you running GuixSD, or Guix on a foreign distro? <ngz>I'm not enough of a daredevil to play with GuixSD ;) <pkill9>try running `export GUIX_GTK3_PATH="$GUIX_PROFILE/lib/gtk-3.0${GUIX_GTK3_PATH:+:}$GUIX_GTK3_PATH"` in shell (obviously replacing $GUIX_PROFILE with yours) and see if it does it <amz3>ng0: you mutated into ngz :D <amz3>ACTION trying to be funny <ngz>pkill9: OK, I'm going to try this (I'm in a middle of a guix pull at the moment). However, is this environment variable new? I re-installed my system (including Guix) recently but didn't have this set before. <ngz>Also, why isn't it in my profile then? <pkill9>dunno, I instaleld somethign one time and it suggested it <ng0>amz3: yeah it was confusing at first <ngz>amz3: ng0 is the hero part of me. <ngz>ACTION is also trying hard to be funny. <pkill9>i just leave it set in my .profile since it's a guix-specific variable and therefore won't interfere with any of my distro's native software <pkill9>also i don't think Guix adds these variables to your guix profile's /etc/profile <pkill9>i have a bunch of suggested ones I added to my ~/.profile which haven't appeared in my Guix's /etc/profile <pkill9>(suggested by Guix after I install certain software) <rekado_>oh… this could be because I played with the grafts module. <mbakke>pkill9: I typically source /.guix-profile/etc/profile in .profile (or /etc/profile.d/) rather than adding them manually. <mbakke>I mean ~/.guix-profile/etc/profile <pkill9>actually it does add those suggested variables to .guix-profile/etc/profile <ngz>But not GUIX_GTK3_PATH, AFAICT <pkill9>i wonder why it doesn't ad GUIX_GTK3_PATH <wigust>davidl: What about “libxft” package? <davidl>wigust: xmonad-x86_64 error while loading shared libraries <davidl>wigust: "..shared libraries: cannot open shared object file libxft.so.2: No such file or directory" <pkill9>hmm dunno why i thought your problem was related to GTK, since i just realised you said it says that when installinga ny package <wigust>davidl: Try to install “libxft” package. <davidl>wigust: I can find it in the gnu store <rekado_>it would only set those variables if you had the gtk+ package installed in your profile. <wigust>davidl: The store should has “libxft” as well as “~/.guix-profile”. <wigust>davidl: “find -L ~/.guix-profile -iname 'libxft*'” should output “libXft.so.2”. <davidl>wigust: .guix-profile doesn't have it so far <wigust>davidl: This means you didn't install “libxft”. What was the command? <davidl>it's a package in /etc/config.scm <davidl>wigust: guix system reconfigure /etc/config.scm <davidl>wigust: I get "shepherd: Evaluating user expression (register-services (primitive-load "/gnu/sto?")). \\n Service: user-homes could not be started. <wigust>davidl: the last one is a different issue not related, i think <davidl>wigust: ok. Should I try installing with guix package -i libxft? <wigust>davidl: What is your “printenv LIBRARY_PATH”? <davidl>wigust: /run/current-system/profile/lib <pkill9>is there a virtual machine available to download in .vdi format? (virtualbox) <davidl>wigust: libxft.so.2 etc exist in /run/current-system/profile/lib as symlinks. <wigust>davidl: I'm not sure, but I think “gcc-toolchain” installed in your user profile doesn't like system profile libraries. If you will install “libxft”, it will hint about exporting “LIBRARY_PATH”. <davidl>wigust: I was able to run a default xmonad-profile before I installed gcc-toolchain. <wigust>davidl: Because the default was compiled during xmonad build, wasn't he? <wigust>davidl: Mixing profiles (system's and user's) could be tricky, especially if they are using differen Guix version (usually it is). <mbakke>pkill9: I believe `qemu-img` can convert from qcow2 to .vdi. Haven't tried it that way though. <davidl>wigust: so if I login as my regular user and update the system to 0.14 with guix pull and system reconfigure /etc/config.scm - would it perhaps be resolved then? <davidl>wigust: thx for your help. Ill have to look at this again another time. <wigust>davidl: np, but you choose a really hard way. Much easier to install xmonad, gcc-toolchain, libxft as a user which will compile xmonad config and use xmonad