<sdb>do any of you have emacs-debbugs installed? <sdb>"By default, Emacs (installed with Guix) “knows” where these packages are placed" this seems no longer to work as my emacs-debbugs installs itself to ~/.guix-profile but this dir is not in any of my paths. <sdb>The manual does not state which variable is supposed to be set automatically... <RetardedOnion>pkill9: i cannot get .profile to work. if i export the variable and restart xfce-panel everything works though. should i put it in my xinitrc? does the xinitrc get called? <pkill9>how do you start the GUI normally? <pkill9>not sure why it's not getting called for you <pkill9>pretend i didn't say 'oh' and 'oh nvm' i was thinking out loud ***mbowcutt_ is now known as mbowcutt
<RetardedOnion>pkill9: nope. does not work. i could add an autostart item that kills the panel, exports the variable and starts the panel again. <pkill9>RetardedOnion: what line did you put in your ~/.profile? <RetardedOnion>export X_XFCE4_LIB_DIRS="$X_XFCE4_LIB_DIRS${X_XFCE4_LIB_DIRS:+:}$HOME/.guix-profile/lib/xfce4" <pkill9>RetardedOnion: do you have a ~/.bash_profile? from searching a little it's suggested that if ~/.bash_profile exists, the shell ignores ~/.profile <pkill9>not sure if it would even matter if the graphical login is used <RetardedOnion>pkill9: i do have one. if i could use guix system (which i cant for some reason) id use zsh though <pkill9>oh that's a common issue lately, you need to run `guix pull` again to fix the lack of `guix system`, due to a recent update <pkill9>anyways try putting that line in the .bash_profile, or just removing the .bash_profile if you're not using it (or just rename it temporarily) <RetardedOnion>if i install tlp, do i need it to have it installed as root? <pkill9>i dunno, i'm not sure how tlp works <RetardedOnion>pkill9: seems like it works as a user. thinkpad specific features are not in the repo though. man i have a lot to do ***mbowcutt_ is now known as mbowcutt
<pkill9>what's the contents of ~/.config/guix for the user you ran `guix pull` as? <pkill9>have you tried logging out and then back in after running `guix pull`? basically the newer guix that will work is in ~/.config/guix/current/bin/guix <pkill9>and maybe logging out then in will set the PATH correctly so your shell uses that one <pkill9>try logging out then in to see if it works then :P <pkill9>sorry idk then, someone else will know though, it's a common issue since a recent update <RetardedOnion>export PATH="/home/lars/.config/guix/current/bin${PATH:+:}$PATH" worked. <RetardedOnion>i know i commented it out. otherwise it wouldnt work. pretend i didnt forget to remove the ;; <g_bor[m]>RetardedOnion: You should use cons* if you are trying to add more elements to the head of a list, like you do it on you filesystem declaration. <g_bor>sneek: later tell RetardedOnion: You should use cons* if you want to add more than one element to the head of a list, like you do it on your file-system configuration. <RetardedOnion>g_bor[m]: thanks. i think i got it and added the uuid as root as well. lets see if i can boot tomorrow <sneek>RetardedOnion, you have 1 message. <sneek>RetardedOnion, g_bor says: You should use cons* if you want to add more than one element to the head of a list, like you do it on your file-system configuration. <g_bor[m]>Ok, I was not sure you are still here, just wanted to make sure you got the message :) good luck! <atw>Hello #guix! An odd thing happened and I'm not sure how: my info appears to be partly in French https://i.imgur.com/zmuly18.png ! I am pretty sure my locale has never been anything but (locale "en_US.utf8") <sneek>atw, ArneBab says: bbad = be back another day <atw>Wish I could help a little more, but I'm far from an expert on this :) <efraim_>Lambda for executing, find-file for getting all the files <efraim_>I'd look for (file) or (dir) in the code <atw>thanks efraim_! I might have found something in python-build-system <atw>Copenhagen_Bram: that looks right! I was just going to say, install it for your user <minieggs40>gged out. Can anyone point me in a good direction? <minieggs40>Puts me back at the gdm login screen. It's feel like the same type of error you get if you have something bogus in your .profile (but the logout occurs when running a guix command rather than on login) <minieggs40>ah silly me. This is a very minimal install of Ubuntu. No xorg <minieggs40>hm, seeing`unexpected Nix daemon error: interrupted by the user` when I `cat syslog | grep 'error'` right after a logout. I'll check what googs has to say <wigust>minieggs40: You could try to run a guix command in ‘screen’ or ‘tmux’ session which will preserve the output of a command. <wigust>After running a command you could check for your environment variables are. XDG_* could make strange things with a login manager. <random_auroras>Should the "boot" partition (/dev/vda1) be mounted for system install or not, given the target is "/dev/vda1" rather than "/boot"? <wigust>random_auroras: it's not required <wigust>random_auroras: you could install guixsd in a single partition on non-efi system <wigust>random_auroras: just unmount it and make sure you didn't specify it in config.scm <random_auroras>Not specified under "file-systems" but specified under "bootloader", correct? <wigust>in your case (target "/dev/vda") <wigust>random_auroras: so yes, don't specified under file-systems because don't need this partition and could delete it <random_auroras>So if I launch it as-is, I'm going to have a partition that's not used? <wigust>random_auroras: it will no be used much either way, e.g. on my non-efi machine the /boot is 8.9Mb <random_auroras>Is there a way to configure root user as disabled and another user as sudoer-admin? <random_auroras>I suspect it may be "no" because it would interfere with guix's processes. <wigust>random_auroras: What do you mean by “interfere with guix's processes”? <g_bor[m]>pkill9: I guess it's in glibc. I will have a look. <pkill9>i thought it would be there also but it isn't <g_bor[m]>efraim_: That was also my second thought, but it does not seem to be there either... <sirmacik>Hi there, I'm trying to install guix and it seems I do everything according to documentation but after successfull install grub cannot find kernel image, here is my config for that section https://pastebin.com/jLHagSQt if anybody can take e look at it and tell me if there's something wrong <g_bor[m]>sirmacik: At first sight this config file is using outdated syntax. What version of guix are you using? <sirmacik>latest from the website and syntax from handbook :/ <g_bor[m]>sirmacik: if you are using a release, this should be okay, but will need modification at the first reconfigure. You can read the latest manual after guix pulling on the new system using info '(guix)'. <sirmacik>stuck at grub trying to find kernel bzImage <roptat>sirmacik: doesn't grub asks for the password to decrypt the disk? <sirmacik>just tells me to load kernel image first <sirmacik>guix system init goes smoothly, no errors reported <g_bor[m]>sirmacik: device should be "/dev/mapper/root". (at least that is how the system tests are run) <g_bor[m]>sirmacik: please report back with the result, as if this turns out to be correct, then we should fix it in the manual. <sirmacik>g_bor: I just need to finish the transfer and log into vpn, 15 mins tops (; <pkill9>how do binaries typically know which paths to search for libraries (without a specified rpath)? <pkill9>so does glibc tell them where to look? <efraim>Looking at strace sometimes I feel like they flail about and somehow manage to figure it out <pkill9>what i'm trying to do is create a /lib64 directoyr which will be a link to a bunch of libraries that a binary might use if you download it <pkill9>but currently doesn't seem to find it <roptat>that's how they know where to find their libraries <pkill9>oh ok, maybe i modified this particular binary <roptat>they look in /gnu/store directly <roptat>at least that's how I think it works <pkill9>hmm one binary i looked at only has $ORIGIN set as rpath <pkill9>annoyingly, running strace doesn't seem to show me where they're looking <pkill9>or maybe i didn't search it right <pkill9>oh actually i found where it's looking <pkill9>it looks in <glibc's store path>/lib <pkill9>I'll just use LD_LIBRARY_PATH then in a wrapper <g_bor[m]>I'm looking around this ldconfig is missing issue, and it seems that we have a few patches that just modify packages to not need ldconfig. So it seems we don't have it. <ng0>so let's say I have something which constitutes a compiler but split over many many files.. is that datadir material? <ng0>I'll just put it over there and hope it continues to be at least as broken as it is now :) <sirmacik>g_bor: I'm waiting for guix system inti to finish <g_bor[m]>sirmacik: I've got to go now, please leave me a message if this works. <g_bor[m]>sirmacik: If you can paste the full config, I would later take a look. Also, you can have a look at gnu/tests/install.scm in the git repository. It might reveal further differences. It contains a full os config and a full installation script for the luks case. <mbakke>sirmacik: I'm not sure if separate /boot is supported. <ng0>you simply mount it? <sirmacik>luks doe not support not having separate boot partition a far as I know <mbakke>sirmacik: GRUB supports full disk encryption, /boot and all. <mbakke>I'd recommend trying without /boot on a separate partition. <mbakke>ng0: Do you have /boot on a separate partition? I haven't tried it, but seem to recall some chatter about it not working. <ng0>mbakke: used to, maybe somewhere on some computer I haven't looked at long <sirmacik>mbakke: now grub asked me for password but later says that /dev/sda2 isnt available despite /dev/mapper/root being declared in config.scm <mbakke>Is this with the 0.14.0 install image? <mbakke>Perhaps GRUB doesn't understand Linux device node names. <mbakke>sirmacik: Can you try to run `cryptsetup luksUUID /dev/sda2` and update the mapped-devices section to read (source (uuid "abcd1234-...")) ? <ng0>can someone tell me what's up with python recently on Guix-on-GuixSD? like, seriously. I can't work with it anymore the way it is <ng0>on Ubuntu or whatever, no issues with Guix. But with GuixSD it's just in a terrible state <ng0>I'm trying the humble and simple task of making it find jinaj2 right now oO <ng0>guix environment.. nope <ng0>with PYTHONPATH set.. nope <apteryx`>fetching the CVE 2017 database seems to tage ages when linting... hm <apteryx`>ng0: did anything change in the way it finds the other modules (PYTHONPATH) ? <ng0>yes. because a long time ago this just used to work <ng0>I'm not doing anything special <apteryx`>there was a discussion to improve things (make python2 and python3 use different env vars, say) and fix virtualenvs too. <sirmacik>got whole guixsd install procedure memorized and I didn't even got it working yet (; <apteryx`>sirmacik: eh, don't dispair! Where do you get stuck? <sirmacik>apteryx`: right now I have grub asking for password but not mounting root, another option (device 'root') doesn't work, now trying system init with (device '/dev/mapper/root') <pkill9>is it possible to remove a 'snippet' from an inherited package's source? <sirmacik>scm-error misc-error and something with Luks Partition <apteryx`>sirmacik: did you mask your UUID on purpose here? I says: (uuid "XXX-XXX") <apteryx`>I doubt "XXX-XXX" identifies your partition :) <apteryx`>I'm not using luks myself (it wasn't yet supported when I installed GuixSD I think -- so I'm double checking using the Guix manual) <aminb>do any of the core devs hang out here? <pkill9>is a guix environment and a guix profile exactly the same functionally? <sirmacik>apteryx`: I did so too, but it looks like handbook version and still won't boot <atw>when writing a build phase, what keys can be used? is it those in a bag? asking because I wrote a phase like (lambda* (#:key version ...) ...) and version appears to be #f <efraim>Inputs, outputs, make-flags, probably a couple of others <apteryx`>atw: See (guix build-systems gnu) for the gnu-build-system. The procedure being passed those args is gnu-build, IIRC. <apteryx`>oh, actually no, it's the `lower' procedure, which uses gnu-build <apteryx`>which returns a bag, so you were right :) <apteryx`>but it is a thin wrapper above gnu-build, and supports all arguments of gnu-build, so the builder itself might be a better reference to know what args are supported. <atw>apteryx`: thanks! I have been poking around in that module a bit to learn the ins and outs of build systems <atw>I might (delete 'haddock) just to make this build faster in dev time <efraim>looks like json-glib for armhf broke on staging after the meson 'zero? system' -> invoke change <mbakke>atw: The version variable is available in the builder, but you have to unquote it. <mbakke>sirmacik: I remember there was a bug regarding the use of "/dev/mapper" paths at some point. My only suggestion is to use a UUID instead of the device path in the (file-system ...) declaration. <mbakke>sirmacik: If you have Guix installed on some other system (not necessarily GuixSD), you can generate an updated installation image easily, perhaps that will help.. <mbakke>sirmacik: Reading the manual, you may get away by just using (device "root") instead of (device "/dev/mapper/root"). <mbakke>PotentialUser-67: Try sourcing `/var/guix/profiles/per-user/root/guix-profile/etc/bash_completion.d/guix`. <fis>Thanks, I will try now. <fis>It works! Thank you very much. <pkill9>please can someone attempt to reproduce my problem <pkill9>just run this and it should unpack kxmlgui-5.47 but instead it unpacks kxmlgui-5.42: <pkill9>wigust-: that's not the command, the one i posted downloads the file into /tmp with a name that guix will use, and then builds kxmlgui with that newer source <pkill9>or atelast it's supposed to, but instead it keeps trying to build with the guix packaged source <pkill9>even though it doesn't say that --with-source had no effect <efraim>"some of debian's smaller build systems now spend SEVERAL DAYS performing the linker phase for these insanely-large binaries." <efraim>pkill9: i normally do 'guix build foo --with-source=/path/to/foo' <pkill9>do you have any idea what's causing the issue efraim? <pkill9>ksmlgui doesn't build, because of an issue that's been fixed in the newer kxmlgui <apteryx`>sirmacik: did you obtain your UUID with the recommend command: `cryptsetup luksUUID /dev/yourdevice' ? <efraim>normally because of patches or snippets, but I don't see one for kxmlgui <wigust->pkill9: --with-source= kinda weird imo. Just edit the recipe directly (you want it after all) <pkill9>wigust-: it does the same thing if i write my own recipe that inherits kxmlgui, but i'll try copying the whole recipe <sirmacik>is there any instruction for generating own installation image? <nomoreiluminatti>while you install guixsd and you want to mount the efi grub which command do you have to write <wigust->pkill9: if you write or inherit recipe, you could play with it in Guile REPL and figure out what's wrong <g_bor[m]>sirmacik: actually there is a method in the manual, see building the installation image, but basically this: guix system disk-image gnu/system/install.scm <sirmacik>right now I'm waiting for system init with uuid mounting to finish <nomoreiluminatti>while you install guixsd and you want to mount the efi grub which command do you have to write <g_bor[m]>sirmacik: yes, I'm trying that now, first with system vm, then if that is ok I will have a look at it on a fresh vm. <sirmacik>I ment to paste url, not the whole file :/ <wigust->nomoreiluminatti: Do you mean EFI partition or what do you mean by “efi grub”? <wigust->nomoreiluminatti: If your EFI partition is ‘/dev/sda1’ then ‘mount /dev/sda1 /boot/efi’ <nomoreiluminatti>how does the configuration template has to look if i want to install a guixsd with an efi based grub and how do i mount an efi based grub <nomoreiluminatti>how does the configuration template has to look if i want to install a guixsd with an efi based grub and how do i mount an efi based grub if my efi partiton is in /dev/sda 1 my root is in /dev/sda2 and my swpa is in /dev/sda3 ? <nomoreiluminatti>how does the configuration template has to look if i want to install a guixsd with an efi based grub and how do i mount an efi based grub if my efi partiton is in /dev/sda 1 my root is in /dev/sda2 and my swpa is in /dev/sda3 and i didn encrypted my root? <nomoreiluminatti>how does the configuration template has to look if i want to install a guixsd with an efi based grub and how do i mount an efi based grub if my efi partiton is in /dev/sda 1 my root is in /dev/sda2 and my swpa is in /dev/sda3 and i didn encrypted my root? <nomoreiluminatti>how does the configuration template has to look if i want to install a guixsd with an efi based grub and how do i mount an efi based grub if my efi partiton is in /dev/sda 1 my root is in /dev/sda2 and my swpa is in /dev/sda3 and i didn encrypted my root? <mbakke>nomoreiluminatti: There is a bug in 0.14 where you have to mount the EFI partition at "/boot/efi", even if your target hard drive is mounted at "/mnt". <g_bor[m]>sirmacik: Did you get anywhere with this? <sirmacik>g_bor: right now grub asks me for password and displays menu, I press enter and then it tries to boot but cannot mount/find root partition <sirmacik>so with entire disc encrypted without separate boot partition works better than before <g_bor[m]>sirmacik: ok, I'm now editing my config, and then will run init here. If it works, I will share my findings. <mbakke>domenkozar: Cachix looks nice (also, fancy seeing you here! ;-)). <domenkozar>I don't know guix ecosystem enough to judge if that's interesting or not <mbakke>domenkozar: I guess it could be useful for third-party repositories/channels. <efraim>nomoreiluminatti: please don't spam <domenkozar>mbakke: is there equivalent of overlays in guix? <mbakke>domenkozar: There is a rather primitive facility called $GUIX_PACKAGE_PATH where Guix looks for package definitions. <mbakke>Channels are still in the planning phase.. <mbakke>domenkozar: How do users of Cachix discover signing keys? <mbakke>E.g. if I publish my builds there, users will need to authorize my key. <domenkozar>that configures Nix to use the binary cache <name>.cachix.org <mbakke>I have a rather popular Chromium package that's still not upstreamed, I'm sure many people would be happy to be able to get substitutes for it. <apteryx`>sirmacik: I can't see any problems with your config... so maybe the aging 0.14 release has indeed a bug, or your partitions have a problem. <apteryx`>Failing to generate a fresh install image yourself, you could try to update guix using 'guix pull' while it's running. There are some bumps going from 0.14 -> current though, so it might be easier/safer to prepare a fresh installation medai. <g_bor[m]>apteryx`: I'm trying to check the release now. If that does its job, we will try to check the partitioning. <apteryx`>ACTION goes back trying to fix gnucash 3.0 <g_bor[m]>sirmacik: Ok, here it works with a no separate boot partition and uuid setup. <g_bor[m]>So I suspect it is something related to partitioning/cryptsetup/mounting. I will share the working config. <g_bor[m]>Can you list for me the commands you executed before system init? <g_bor[m]>sirmacik: One more thing, could ou try to remove the dependencies field? <g_bor[m]>sirmacik: bug #31889 is something related to that. The dependency is already expressed by the device "/dev/mapper/..." line. <sirmacik>setting up internet connection, cfdisk, creating partition with bootable flag, cryptsetup luksformat, cryptsetup luksopen, mkfs.ext4, mount to /mnt, herd start cow-store /mnt, copying or editing config.scm, guix system init <sirmacik>g_bor: even with dependency line removed problem presists :/ <ng0>I think we should make guix info $foo return info guix $foo... sometimes I type this :D like guix push or guix log xD <mbakke>sirmacik: What hard drive do you have? Could it be that the initrd lacks a driver for it? <sirmacik>its vmware virtual machine, they generally work with open-vm-drivers <sirmacik>boots and mounts without a problem on guix live <mbakke>Ah, that explains it! There is an open bug for it. <mbakke>sirmacik: You may be able to work around it by changing which disk controller to emulate. <mbakke>sirmacik: At least you're a pro at doing Guix installs now, might as well jump on real hardware ;-) <sirmacik>just need to try it out for a day or two and that's the general idea (; <sirmacik>but I also wanted to run it on few test servers to streamline the setup, any tip on which disc controler might be the right one? <mbakke>Not a great first impression, I'm sure :D <mbakke>sirmacik: Sorry, I haven't worked with VMware in a decade. <mbakke>If you do find a working one, could you update the bug report? :-) <sirmacik>at my institute unfortunately most of the infrastructure is built upon vmware <pkill9>it seems 'ktextwidgets' and 'kglobalaccel' as inputs cause guix to start building guix version of kxmlgui when you try to build newer verison of kxmlgui, weird <sirmacik>need to take few hours of sleep and I'll get back to it tomorrow, thanks for todays help guys! <pkill9>that's enoughdebugging for today <sirmacik>mbakke: as for impression I'm quite determined to get it running and production ready (; <sirmacik>mbakke: thx, I'll try to determine which one is missing <mbakke>sirmacik: According to the bug report, you need mptbase, mptscsih, mptspi :-) <sirmacik>ah, just another situation pointing out that I should get some sleep (; <sirmacik>added modules, trying system init one last time for today <sneek>atw, ArneBab says: ice-9 are core tools provided by Guile "as a crystal seed" for programs. <sneek>atw, ArneBab says: SRFIs cross-implementation features. They are collected at https://srfi.schemers.org/ and are likely to work on multiple scheme implementations, not only in Guile. <mbakke>sirmacik: Wow, that's new! Can you share a screenshot when available? Good night! 