IRC channel logs


back to list of logs

<pkill9>dw i'll check the existing python packages
<rekado_>pkill9: you don’t. You can override the Python version in the “arguments” field.
<pkill9>ah ok
<rekado_>but only if it’s a python-2 only package.
<rekado_>otherwise we have a procedure that generates a python-2 variant.
<wigust_>pkill9: Just curios, Doesn't "guix refresh" work for update the local package definitions?
<rekado_>davidpgil[m], pkill9: please only add this line to .bash_profile to avoid problems when using sub-shells (like those spawned by guix environment).
<rekado_>wigust_, pkill9: “guix refresh -u” does indeed update the sources, but you need to have a writeable copy of them.
<rekado_>i.e. you’d use it with a git checkout and “pre-inst-env”
<davidpgil[m]>rekado_: I found this answer after I asked in this forum. works excellent. May be trisquel exclusive, not sure.
<pkill9>hmm looks like it did install correct version afterall
<pkill9>i'll open an environmental shell and see if i cna import the pythong package
<pkill9>hmm, it decided to see python-attrs after running `guix environment <some packages>` a few times
<jaidmin>hi, im having troubles with the icon themes and the shell themes for gnome
<jaidmin>the dont show up in the tweak tool, anything that im missing?
<jaidmin>do i have to manually put a symlink somewhere? sorry if its obvious but im pretty new to guixsd
<jaidmin>so they get picked up by the tool when i manually put them in ~/.themes
<jaidmin>but not when i install the packages
<pkill9>is there a way to roll back a 'guix refresh -u'?
<pkill9>if not then i think i borked my install lol, i'd rather not recompile everything :P
<wigust_>pkill9: "refresh -u" changes files not in guix store, so a version-control program or a backup before "refresh"
<wigust_>pkill9: similar situation with user configuration files which guix doesn't control by default
<pkill9>anyone running proprietary nvidia drivers here?
<efraim>Good morning!
<efraim>hmm, not sure I really want to independantly package each of lumina's desktop utilities
<civodul>Hello Guix!
<rekado_>civodul: hi!
<civodul>how's everything?
<civodul>mb[m]1: should we merge the openssl upgrade in core-updates and build the whole set of packages?
<rekado_>I’m still having trouble with a module cycle, unfortunately.
<rekado_>adding “r-igraph” from (gnu packages graph) to (gnu packages cran) breaks everything.
<rekado_>civodul: when we rebuild almost everything anyway could we also merge the HTTP/2 patch for curl?
<CharlieBrown>Guix was ported to the Hurd. How do I run a GuixSD system based on Hurd? I have a 32-bit machine.
<civodul>rekado_: for cURL i think it's still time, though it's good to check with mb[m]1
<civodul>rekado_: re cycle, does r-igraph inherit from something?
<efraim>looking good on the lumina-1.4.0-p1 patch, I might actually have it ready for a patch submission this version
<rekado_>civodul: no, it doesn’t. I’m checking its inputs now.
<efraim>right now i'm test building mesa with vulkan on aarch64 on core-updates
<civodul>CharlieBrown: GuixSD on GNU/Hurd doesn't exist yet, though rennes published something that's a first step in this direction
<civodul>see the guix-devel archive from last week or two weeks ago
<efraim>vulkan passes -msse2, so definately intel only
<wingo>good morning guixfolk
<civodul>heya wingo!
<danialbehzadi>Hey. I just installed guixsd 14 on Qemu, but I just got a guile prompt after booting. What should I do next?
<civodul>danialbehzadi: could you report the error that appears above the Guile prompt?
<civodul>it may be that it failed to boot your root partition, for instance
<danialbehzadi>Error: In procedure scm-error
<danialbehzadi>Error: LUKS partition not found #vu8
<civodul>so it looks like it could not find a LUKS partition by the UUID you gave
<civodul>is it the root partition that's encrypted?
<danialbehzadi>No. There is no LUKS partition
<civodul>which config file did you use as a starting point?
<civodul>among those at
<civodul>the GNOME/Xfce example expects a LUKS-encrypted root partition
<danialbehzadi>civodul: This is my config:
<danialbehzadi>Oh, I see. Thats `type luks-device-mapping`
<danialbehzadi>Can I do something to fix it witout fomat and encrypt whole disk?
<civodul>yes, so you need to either remove that (mapped-devices ...) form, or properly format your root partition as LUKS
<civodul>either way you have to reinstall
<wigust_>danialbehzadi: Also (dependencies mapped-devices) probably needs to be removed.
<danialbehzadi>Thnaks. I'll try it ;)
<g_bor>I've been thinking for a while how to make an installation which is compatible the the guix philosophy...
<g_bor>It seems to me, that an installation file, similar to the current operating-system-configuration should be made.
<g_bor>Anyone working in that direction currently?
<g_bor>I'm not very familiar with the current installer effort.
<mb[m]1>civodul: openssl upgrade?
<mb[m]1>With the recent nurses rebuild, we might as well take Aruns patches and the tar update.
<efraim>I'm pretty sure qt5.10 depends on a newer sqlite than we have in master
<civodul>heya mb[m]1!
<civodul>mb[m]1: yeah, mark_weaver pushed a new openssl replacement in master
<pkill9>hi, how can i get software installed with Guix to use my system's Nvidia proprietary graphics driver?
<mb[m]1>Oh, okay. Yeah let's merge and ungraft it. I can do so tonight if no one beats me.
<pkill9>otherwise they fail with the same error: `libGL error: failed to load driver: swrast`
<mb[m]1>civodul: Are core-updates evaluations working on Hydra now?
<wigust_>pkill9: People on #guix don't help with proprietary stuff usually.
<civodul>mb[m]1: good question, maybe not
<civodul>it's kinda weird
<pkill9>wigust_: i thought so yea, i thought I'd ask incase soemone had the answer
<civodul>ACTION tries
<civodul>rekado_: FOSDEM main track talk rejected
<civodul>another year maybe!
<pkill9>civodul: they rejected a talk onguix?
<pkill9>on Guix*
<rekado_>civodul: oh, too bad!
<civodul>oh well
<pkill9>i keep seeing people accidentally call guix 'git'
<efraim>i'm going to test mesa with llvm for aarch64 on core-updates, see if we can reenable that for swrast graphics
<efraim>i keep on forgetting perl when I test debootstrap
<civodul>efraim: when you test debootstrap?
<civodul>what are you doing with debootstrap? :-)
<efraim>It seemed like it might be a helpful package to have
<efraim>I'm going to have to keep,on looking at apt-cacher-ng to see what the minimum needed to make it run is
<mb[m]1>Ooh, core-updates evaluation pending. At least it didn't fail immediately :)
<civodul>mb[m]1: i don't think it failed immediately previously
<shiranaihito>efraim btw, do i need to worry about "sector size" (or other similar stuff) when partitioning a disk for GuixSD?
<efraim>Shouldn't need to
<efraim>I normally go with a swapfile over a swap partition unless I'm on btrfs
<shiranaihito>efraim does this look good? - now there's a 1MB partition meant for GRUB
<shiranaihito>why? :) a swap partition seems easier to me at least
<efraim>Its easier to grow and shrink swapfiles
<shiranaihito>efraim ah, well that does make sense
<shiranaihito>"Information: You may need to update /etc/fstab." <-- how do i do that?
<shiranaihito>(i ran "parted /dev/sda set 1 bios_grub on" to get that message)
<mb[m]1>civodul: It's alive!
<civodul>i can't believe it!
<shiranaihito>yay, i got GuixSD running (in a VM) for the first time :)
<g_bor>Hello guix!
<g_bor>I think I need some clarification.
<g_bor>I've seen when we have svn-fetch as a method, then sometimes a filename is specified.
<g_bor>What is that good for?
<rekado_>do you mean the “file-name” field?
<rekado_>that’s to give the store item a nicer name.
<g_bor>rekado_:yes, exactly, thanks
<jaidmin>anyone got network-manager-openvpn to work? I installed it but I still cant use openvpn in networkmanager gnome
<pkill9>is guixSD technically a stateless operating system?
<nextos>Hi, im trying to install guixsd on a EFI machine (an old intel nuc), but it fails with grub-install: error: failed to get canonical path of `/boot/efi'.
<nextos>I'm probably doing something silly
<nextos>My bootloader declaration is copied from Guix docs
<nextos> (bootloader (bootloader-configuration
<nextos> (bootloader grub-efi-bootloader)
<nextos> (target "/boot/efi")))
<nextos>I also have an appropriate filesystem declaration for the EFI partition
<nextos> (file-system
<nextos> (device (uuid "1234-ABCD" 'fat))
<nextos> (title 'uuid)
<nextos> (mount-point "/boot/efi")
<nextos> (type "vfat"))
<nextos>with my uuid instead
<nextos>I also tried mounting the EFI partition under /mnt/boot/efi before running guix system init config.scm /mnt
<wigust_>nextos: Try (device (uuid "1234-ABCD" 'fat)) -> (device (uuid "1234-ABCD"))
<nextos>wigust_: thanks. With that change, guix system init refuses to even start: config.scm: error: invalid UUID
<nextos>however my UUID matches the UUID listed in blkid for
<nextos>that disk
<nextos>shall i use a PARTUUID instead?
<wigust_>nextos: I use "/dev/sda1" for efi partition
<wigust_>don't remember partuuid vs uuid
<nextos>wigust_: ok
<nextos>wigust_: it still fails. Do you mind posting the relevant lines from your config.scm AND clarifying where you mount the EFI partition before running guix system init
<nextos>Or just anyone else, if anyone has a EFI configuration running (on x86_64), do you mind posting your config.scm and the specific steps you carried out during install (eg format or mount the partition)
<nextos>efraim: thanks
<nextos>efraim: the url you sent times out from different locations, seems unreachable
<efraim>Its hosted on the RPI sitting behind me
<nextos>efraim: no worries, thanks again, it works now
<efraim>missed a couple of CVEs against rsync
<efraim>i'm patching them now
<bebarker>just learned about guix and am curious which repo and directory is the entrypoint for the guile package definitions for all guix packages?
<roptat>all package definitions are in the guix repo under gnu/packages
<ArneBab>did you see this?
<bebarker>roptat - thanks!