IRC channel logs


back to list of logs

<lechner>Hi, is sudo guix system reconfigure ./operating-system.scm the right command? I'm getting guix system: error: failed to load './operating-system.scm': No such file or directory
<jakef>does the file operating-system.scm exist in your current working directory?
<jakef>it's the command to update your system according to what's in that file
<wargreymon>Hi, im learning about gexp. I tried "(file-like? obj)", dont know what to type in, the function keeps output #f, what "obj" should it be?
<lechner>jakef / yes, that's the weird thing! it does exist. i'm only using that command because guix deploy stopped working locally
<db48x>lechner: when did it stop working? what were you doing at the time?
<db48x>for example, did you change sudo’s configuration? your config can mandate that commands are run in a particular working directory, which would explain why guix cannot open the file
<db48x>of course this is pretty unlikely
<db48x>running `sudo pwd` might shed some light on the matter
<civodul>Hello Guix!
<thaenz>Good morning everyone! :)
<abbe_>hi! anyone here using guix on aarch64 VMs ?
<abbe_>while guix qcow2 image built with linux-libre 6.7.10 works, 'guix system reconfigure' with kernel = linux-6.7.12, or linux-6.8 doesn't :(
<civodul>abbe_: hi! how does it not work, more precisely? :-)
<abbe_>sorry, after selecting the menu entry corresponding to the generation from the GRUB menu, there is nothing on the display, as I could see in the provided VNC display.
<abbe_>the original generation (present in the disk image) continues to work fine
<abbe_>i can share the configurations shortly of the disk image, as well as diff which fails to work for me
<civodul>abbe_: could it be a problem with a LUKS-encrypted device?
<peanuts_>"Failure to open LUKS devices from a Shepherd service"
<peanuts_>"(use-modules (gnu) (gnu packages emacs))(use-service-modules n -"
<peanuts_>"channels.scm -"
<abbe_>'guix time-machine -C channels.scm -- system image --image-type=qcow2 --image-size=5G config.scm' is how I created the disk image.
<civodul>config looks legit to me
<civodul>did you try running the image in QEMU, to get more info?
<civodul>(on x86_64)
<abbe_>the image boots fine
<abbe_>when I run guix system reconfigure from within the booted image (in a VM):
<peanuts_>"(use-modules (gnu) (gnu packages emacs) (gnu packages li -"
<abbe_>grub stuff:
<peanuts_>"guix% grep -C2 -F linux /boot/grub/grub.cfgmenuentry "GNU with Linux-Libre 6.7 -"
<abbe_>thanks for your help, btw!
<civodul>abbe_: i’m not sure i got it right: the image built with ‘guix system image’ boots fine, but reconfiguring the machine on the same commit doesn’t? is that correct?
<abbe_>almost, when reconfiguring the machine, I'm not using same guix commit (in channels.scm), as the point was to get newer packages, but I guess I can try rebuilding with same channel as the one used in image.
<civodul>yes, if you could built an image with the same commit that you were attempting to reconfigure to
<civodul>and then test it in QEMU, that’d be great
<civodul>you could send the channels.scm + minimal config.scm file to bug-guix if it’s reproducible
<civodul>hey futurile!
<civodul>anyone played with cloud-init in Guix System?
<civodul>looking at it seems to be very FHSy-centric
<peanuts_>"Config Drive cloud-init 22.1 documentation"
<zilti>I am trying to pack a program depending on OpenGL. It seems the package I have to depend on for that is libglvnd, but... While, yes, the program builds with that dependency, trying to run it results in it stopping immediately with an "EGL_EXT_platform_base not supported" error. Am I depending on the wrong package after all? Or is there something additional I need?
<futurile>civodul: not on guix - but generally it's very good - works well as a 'system' on the cloud providers
<abbe_>civodul: figured it, it turns out after removing (options "noatime") from the root filesystem configuration, it worked fine with the latest kernel. apparently that filesystem option was not present in my image configuration
<civodul>abbe_: oh, interesting
<civodul>i wonder how that’d prevent it from booting though?
<civodul>oh i see
<civodul>abbe_: “noatime” is not a valid *option* string, it should be a *flag* instead
<peanuts_>"File Systems (GNU Guix Reference Manual)"
<abbe_>I see, thanks!
<nij->Can guix the package manager work on macos as of 2024?
<blum>To dual boot guix i just install install guix to an empty partition and then select that boot device from bios?
<blum>Or are there any more complications i need to be aware of when dual booting windows and guix?
<db48x>pretty much. the way it works is that every OS puts a “bootloader” into the EFI boot partition
<db48x>then the UEFI “BIOS” gives you a boot menu that lets you choose from them
<attila_lendvai>blum, there's nothing specific to guix in dual booting. make sure you're booting in EFI mode, not the old bios bootsector stuff. then guix will put its boot files to the EFI partition, next to the other OS-es.
<blum>Thanks guys. I'll try it after dinner :^)
<zilti>nij-: unfortunately not, no... Which is why I now have nix installed on top of guix, because we moved away from docker and needed a direnv-based dependency solution
<nij->zilti Thanks for getting back! Are you on macos? And you mean you have nix over guix over macos?
<zilti>nij-: No, I'm on Guix-the-OS and have nix-the-package-manager on top of it for use with direnv. We had to do this because there is no guix on macOS, and my boss and a coworker are on macOS. I think the reasoning for not porting it to mac is "Guix is tailormade for Linux and Hurd". And in practice the project doesn't have the manpower to do it, too. Guix is much smaller than Nix.
<nij->Ah, I see. Since you are collaborating with someone using macOS, you need to adjust to their ways (i.e. NIX)?
<zilti>nij-: Not at all, no - I only had to add the nix-service to my config.scm. Everything else happens automatically via direnv.
<nij->got it
<Guest50>whelp: can't make service, doesn't work and throw : guix home: error: no target of type 'system' for service 'profile
<Guest50> (services
<Guest50>  (list
<Guest50>   (simple-service
<Guest50>    'arei shepherd-root-service-type
<Guest50>well no luck then, doesn't work making simple services
<rekado>I'm getting a weird error on our cluster
<rekado>whenever I want to download any sources with "guix build": In procedure open-file: Permission denied: "/etc/ssl/certs/localhost.crt"
<rekado>not unexpected because /etc/ssl/certs/localhost.crt exists with 600 permissions for root only.
<rekado>the weird thing is that /etc/ssl/certs/localhost.crt exists on this machine where "guix" is run, but not on the machine where "guix-daemon" runs (we use GUIX_DAEMON_SOCKET)
<rekado>why should the existence of this file on the current machine matter at all?
<rekado>I would have expected the download to happen on the node where the daemon runs
<rekado>wget doesn't care about the bad permissions on that file and just downloads it.
<rekado>I don't see why Guix looks at /etc/ssl at all.
<rekado>the command I used is "guix shell bedtools --with-source=bedtools= --with-c-toolchain=bedtools=gcc-toolchain@9", which forces Guix to download sources.
<blum>dual booting question: i installed windows on the computer leaving one partition for guix. Now i'm installing guix. How should I partition the guix drive? Should i make an EFI system partition(drive with ESP flag on) and another partition for the guix root, or do i reuse the windows EFI syste partition?
<rekado>ACTION doesn't know
<blum>ight im gonna go with making a second EFI partition. that option seems less likely to brick my (girlfriend's) windows
<user_oreloznog>blum: hi! Don't know if it's relevant, but I managed to dual-boot (no efi) in 2019
<blum>user_oreloznog: thanks will read!
<blum>user_oreloznog: Gotcha, so then guix owns grub and debian is a boot option in grub?
<user_oreloznog>I don't really remember... Something like this
<dinosauro-trex>hello i've installed guix repo on trisquel gnu linux, can someone help me to install icecat?
<user_oreloznog>blum: I think you're right
<dinosauro-trex>I installed guix from synaptic, then I typed on terminal "guix install icecat" and it installed icecat 78.10.0-guix0-preview1
<dinosauro-trex>but I don't know how to open it
<Kolev>dinosauro-trex: Maybe try logging out and back in.
<dinosauro-trex>okk i'll try
<dinosauro-trex>I restarted the pc
<rekado>sneek: later tell dinosauro-trex Environment variables govern where your shell (or graphical environment) looks for executables when given a command. In a shell session that's the PATH variable. You can set it to the default location where Guix installs things with `export GUIX_PROFILE=$HOME/.guix-profile; source $GUIX_PROFILE/etc/profile`
<dinosauro-trex>but now I see a lot of squares
<sneek>Welcome back dinosauro-trex, you have 1 message!
<sneek>dinosauro-trex, rekado says: Environment variables govern where your shell (or graphical environment) looks for executables when given a command. In a shell session that's the PATH variable. You can set it to the default location where Guix installs things with `export GUIX_PROFILE=$HOME/.guix-profile; source $GUIX_PROFILE/etc/profile`
<rekado>"guix install anything" links package contents to ~/.guix-profile by default
<dinosauro-trex>now I have another problem anyway
<RavenJoad>Did Guix change something about how custom Linux configurations get applied? I am getting "Mismatching configurations in .config and arch/x86/configs/guix_defconfig (("CONFIG_DEVPTS_MULTIPLE_INSTANCES" (#f "y")))" when rebuilding my system.
<apteryx>no, but perhaps new dependencies got added in the kernel
<apteryx>check with menuconfig in doubt
<RavenJoad>apteryx: You mean clone the kernel and run "make menuconfig"? Or does Guix have a set of tooling for that?
<RavenJoad>Because it looks like this mismatch is coming from Guix rather than Linux itself.
<RavenJoad>Ok, looking at (guix build kconfig), it _is_ Guix throwing the error, because the config is invalid, somehow.
<apteryx>I've found the best way to use customize-linux is to produce your own defconfig file
<apteryx>not sure what you are trying to do exaclty
<apteryx>RavenJoad: you can use 'guix build -S linux-libre' to get the kernel sources to run 'make menuconfig' from
<RavenJoad>It is not super important, but I previously had CONFIG_DEVPTS_MULTIPLE_INSTANCES enabled for "isolated PTY namespaces (eg: in containers)". I am too lazy to carry around multiple defconfig files for my different machines. I may just axe the config option.
<cdo256`>Don't you hate it when you're writing a package definition, you get down to the bottom to write the license and realize that the author never got around to adding a license to the project.
<apteryx>ACTION fixed gdk-pixbuf on core-updates
<rekado>cdo256`: yeah, happened to me a couple of times
<blum>Installation working alongside windows! can boot into guix and windows from bios which is good enough for me :D
<blum>first time running guix on hardware :DD
<Kolev>blum: Nice!
<blum>yeah real nice
<Kolev>I'm shocked Windows let you install GNU/Linux on the same disk as Windows.
<blum>just gotta pull over my home config from the blumbox and i should be good to go
<blum>oh and also need a bluetooth controller. any recommendations for something light on xfce?
<blum>thx ill try that
<blum>Kolev: wasn't as problematic as i thought it'd be :^D
<vagrantc>i notice a fair number of python packages call "pytest -vv" ... wonder if it would make sense to make that the default?
<freakingpenguin>Getting some slow substitute updating from bordeaux. Is this normal or is it a problem on my end?
<cbaines>freakingpenguin, feel free to share/email the relevant details
<cbaines>in particular, it's good to know what file you're downloading, what highest speeds you get when downloading things, and roughly where you are in the world
<freakingpenguin>cbaines: Seems to have been transient, cleared up after 15 minutes.
<cbaines>ok, bayfront (the machine that serves is under high load at the moment, so that might have contributed
<freakingpenguin>It was a system reconfigure, most packages were actually fetched from a lan cuirass server. US. Just the initial "updating substitutes from..." was slow.
<freakingpenguin>cbaines: Thanks for checking!
<blum>I'm having these weird instant random shutdowns on my laptop. It tends to occur when im doing something intensive such as guix pull or guix package -u. it also occured on debian though. Tends to occur when laptop is not plugged into power but also sometimes when it is plugged into power.
<blum>I've been grepping through /var/log, but not found the cause yet. anyone have some clues?
<ieure>blum, Sounds like the CPU got hot and went into thermal shutdown. My ThinkPad X230 did that a while back, I replaced the thermal paste and it stopped.
<freakingpenguin>Could also be memory issues, memtest86+ could help but that takes a while
<freakingpenguin>CPU is more likely
<blum>oop there it crashed again. ran guix system reconf -> crash -> guix system reconfg -> crash
<blum>ieure: yeah i think so too
<blum>made a lil platform for it for more air flow
<blum>doesn't rly happen on windows doe afaik
<blum>might be just that my gf doesnt compile much software
<blum>freakingpenguin: i'll try memtest too thank u
<blum>ieure: do u know a way to debug or mitigate that in software
<blum>downclocking perhaps
<Kolev>blum: Overheating?
<blum>Kolev: i think so
<ieure>blum, You could maybe make the fan run faster all the time, and/or underclock (tlp can do this). But those are hacky workarounds that aren't solving the problem, so I dislike them.
<ss2>What would actually solve the problem would be to clean it in the first place. If it is more than five years old and has been tossed around, then it will collect dust.
<ss2>blum: install lm-sensors to your system profile and run sensors to see the temps of your system.
<blum>thx will do. let it cool of and trying reconfigure now.i opened the screen and lifted it such that there's nothing covering the fans. i don't wan to cancel reconfigure process. So i gotta wait until it crashed/reboots before tyring them. Update: crashed again
<blum>wait lemme see if i find an ice pack
<ss2>Place on top of a stone plate.
<blum>well it's 30 degres hree so don't have any cold stone unfortunately
<ss2>You could also reduce the clock of your CPU in your bios or select everything in powersave mode. It will slow things down, but it prolong the crashes a bit.
<blum>> System performance mode: intelligent cooling
<ss2>Anything other than performance will do.
<blum>is it possible to tell guix to use a fixed number of threads?
<ss2>yes, with --cores
<blum>i'll try w 2 cores then.
<blum>set it to battery saving mode
<blum>it says it reduces fan speed too :/
<blum>well intelligent cooling has not been working well and final option was extreme
<blum>extreme perf*
<ieure>blum, Is there a reason why you want to solve this in software instead of repasting the CPU? Stopgap until you get some decent paste?
<blum>ieure: yeah i want to use the laptop
<blum>and cant finish guix installation w/o
<blum>and since it's not my computer i'd prefer not to open it up :/
<blum>i'll turn off the screen, should reduce heat a bit
<blum>hmm. since i dont need the screen i can flip the computer upside down, so that the fans face into open air.
<ss2>Got a fan in the room to blow some air to it?
<blum>ss2: when you say so
<blum>screen off, battery mode, house fan cooling... let's see if this takes me through guix sys reconf
<blum>still shut down, but even though not hot
<futurile>blum: sounds like something else going on .. I would get the memory test out that someone mentioned
<blum>on it :D
<dokma>What is the proper way of compiling and linking with packages installed via guix?
<futurile>dokma: preferred is going to be to create a package, so that you can reference all the libs correctly
<futurile>dokma: a local package for your own use
<rekado>dokma: if you don't want to build a proper package, though, it's really not too different from how you'd do it on any other distro.
<rekado>though you don't have global locations and instead rely on environment variables.
<dokma>that sounds totally strange to me. I am trying to link libguile into my project. Are you telling me that the proper way to do that is to create a guix package?
<rekado>for example, you'd enter a "guix shell" with gcc-toolchain and all packages you need, and then compile things the usual way
<rekado>creating a Guix package (locally) is preferred, because you can be sure that you won't end up with something that links to libraries that are later considered garbage and are removed from your system.
<dokma>I use CMake for building my project. I expected guix would be able to spit out some paths for me and be done
<rekado>what do you mean by spit out some paths?
<rekado>you can do "guix shell cmake gcc-toolchain guile ..." and then run cmake inside that environment
<rekado>this should work just fine.
<dokma>Something along the lines of executing some guix command to tell me where libguile.h and are located
<rekado>Guix sets environment variables that tell the toolchain where to look.
<dokma>So my cmake project is built by just running make. In order for me to be able to link stuff installed by guix, cmake needs to be able to find the headers and libraries no?
<rekado>so run the cmake configuration step and make inside a "guix shell" where the libraries you need are made available
<dokma>interesting... totally unlike anything I've used so far
<dokma>rekado: tx
<blum>Tested memory with memtester with 6gb (i have 8gb ram), and didnt find any mem problems. not quite memtest86+, will try that later
<blum>running guix with one core now, unfortunatley it started building the kernel
<abralek>Hi Guix
<blum>abralek: waddup