IRC channel logs
2026-06-15.log
back to list of logs
<Saorsa> does anyone here happen to know whether the `(service home-service-type)` can be used to embed a home configuration in the system configuration and have it be built by guix system reconfigure? <ieure>Saorsa, It cannot, you're confusing it with `guix-home-service-type'. <ieure>Saorsa, See the manual, `(guix) Guix Services' for detail. <Saorsa>Ah, can `guix-home-service-type` achieve that functionality then? <eikcaz>hmm. Why on earth would /run/user not be populated? <apteryx>is codeberg-cli useful to have in manifest.scm? <apteryx>hm, 'make check TESTS=tests/guix-environment.sh' is prone to hang here <apteryx>hm, never mind, it may be caused by the change under review <apteryx>(the environment variable); and where is this documented? <Saorsa>Cheers for the previous help, think I managed to address the issue I was having. <switchy>apteryx: that's a terminfo(5) thing, as part of ncurses? the toe -a command prints the database entries <Saorsa>Turns out that I was stuck on a pinned version which didn't have the functionality I was after. <apteryx>I was trying to come up with a useful environment variable to set in a 'guix shell' example (the current one runs mpirun) <apteryx>i guess i'll go with OMP_NUM_THREADS=16 <switchy>yeah, I think that's a better choice than TERM which is pretty messy.. <apteryx>pushed the change as d11615f609d; we now use -E SOMETHING=VALUE with guix shell to assign it directly, thanks Thomas! <crinklywrappr>no. i see `guix locate: error: file 'clear' not found in database '/home/crinklywrappr/.cache/guix/locate/db.sqlite' <Saorsa>Only thing now that needs fixing is my swapfile not activating on boot. <crinklywrappr>it's more convenient than `ls -l $(which <command>)`, and inspecting the symlink <Saorsa>Hey, when setting up a swap file is there anything else in particular that you need to do besides what is specified in the manual? <Saorsa>Just that my swap file isn't being initialized after booting despite having the options set for it in my configuration. <robin>node "Keyboard Layout and Networking and Partitioning", lots of topics :) <robin>ah, just the basics for each <robin>hm, dd shouldn't be needed anymore <Saorsa>Say, what is the best way to update your system with Guix? <Saorsa>Just that there seems to be a few ways to about it from unsetting and resetting your profile to using a guix command. <robin>Saorsa (Saorsa60 ?), hm what filesystem are you using? (btrfs is different) <robin>(hmm my firefox is dying, maybe i finally *really* have too many open tabs...) <Saorsa60>I've pretty much done everything from following the Arch Wiki to consulting an LLM at this point. <Saorsa60>For whatever reason, my system doesn't seem to activate swapon at boot. <robin>ACTION checks their btrfs configuration <Saorsa60>I've got my configuration files uploaded on to GitHub for troubleshooting, so I'll shoot you a link over to it. <robin>thanks, that is always helpful :) <Saorsa60>Here's where all my filesystem stuff lives. <robin>the confusing thing about the manual is that swapon/swapoff are for interactive testing, and the boot-time config is different (the commands are only documented for installation-time, to make sure it works i guess, but for ordinary use it's an operating-system clause) <Saorsa60>You'll have to forgive the formatting, I've only just transitioned over to Guix from NixOS. <efraim>you might need some sort of dependency management thing to make the swap target depend on the mount point <identity>Saorsa60: you can try ‹guix style -f hardware-configuration.scm› and see if the output is worse than what you started with <robin>oic, Lisp/Guix/System/Configuration/configuration.scm has the operating-system record <Saorsa60>I'm definitely doing that for the rest of my files. <robin>(that maybe helps my in-place package updater subproject) <identity>i would also recommend doing (define the-disk (uuid "58e043e1-c58a-4d6d-9d7d-e41baf92dc7f" 'btrfs)) or some such instead of writing it 50 times for every sub-volume <Saorsa60>Haha, yeah. I'm sure a lot could be done to cut down on the recursion. I'll consider that once I've got the swapfile working. <robin>Saorsa60, btrfs handles swapfiles differently for some reason, so btrfs subvolume create /var/swap (or so) and btrfs filesystem mkswap --size 4g --uuid clear /var/swap/swapfile <robin>then it should work automatically after reconfigure+reboot <robin>(not sure how to list swap status, mount sounds wrong) <Saorsa60>I've already tried using the swapfile at the root of the filesystem, but that didn't seem to work either and I wasn't really wanting to remove the filesystem compression on it. Not sue whether create it under var will do much different, but I'll give it a shot. <identity>Saorsa60: does it really have to be a swapfile? they just create problems for no reason in particular, i think <identity>just create a swap partition and be done with it <Saorsa60>Only reason I haven't so far is because pretty much all the space in my drive is allocated already, so I'd have to resize it to make a traditional swap partition unless it can be done through btrfs subvolumes as well. <robin>(i like the naming convention, titlecased folders are pretty common in CL projects where pathnames are semi-arbitrary, nontrivial for guile though) <robin>Saorsa60, yeah, it has to be in a separate subvolume <robin>(or...actually i'm not sure i attempted it without) <identity>it does not have to be, but it is a bad idea <robin>oh, no snapshots of volumes with swapfiles according to archwiki <robin>ACTION should set up backups that get updated <identity>‹it›, which is a bad idea, is a swapfile residing on a subvolume with other stuff <robin>yes, i see archwiki [[Btrfs]] has a number of other constraints too (no raid, no multidevice, etc., i also don't love the idea of storing a ram snapshot on a working volume, though swap partitions are probably less popular these days) <JD93>For a Python package's source which is better: GitHub or PyPI? `(origin (method git-fetch) (uri (git-reference ...)) ...)` or `(origin (method url-fetch) (uri (pypi-uri ...)) ...)`? <robin>(i should check if guix has unofficial wikis, some years ago the consensus was that info manuals are enough, which likely remains the consensus) <robin>(moving my rambling to #systemcrafters) <robin>apteryx, hmm i see some mediawiki userboxes at the bottom that i believe some wikipedians might like :) <identity>«This user herds their GNU with Guix System.» <adanska>is there any possibility of downloading all of the irc logs? logs.guix.gnu.org's search function seems incomplete.. it does not show me more recent messages if i try to search for my nickname, for example <adanska>i just want to be able to grep through it to try and find some old conversations i had here <adanska>does anyone know who's in charge of the irc logging? <apteryx>adanska: I use my own bouncer (pounce-service-type), hosted on a Guix System VPS <apteryx>and let ERC do the logging so in theory I have everything I care about saved under ~/log/. <adanska>apteryx, thats a good idea.. though i want previous logs though i will do something like this in the future <JD93>For a Python package's source which is better: GitHub or PyPI? <trev>morning futurile. is there a recording of yesterday's meeting? <futurile>trev: tell me what you think when you've watched it ;-) <trev>futurile: it was good, i listened while cooking lol <futurile>trev: probably a good use of time, I'm having to listen to myself to write the minutes (ugh!) <mdioc>Hey everyone. I have been using guix system for a few months now and really enjoying it. However, I am apparently too dense to get unattended upgrades working. For some background, my system config is split into a shared base config and then inheriting that in my individual systems. My configs are stored in a custom channel I use (not sure if this is the right way to do this) and I cannot figure out how to add my cu <mdioc>stom channel to unattended upgrades service type. I see 'channels' but no matter how I try to append to %default-channels, I seem to get an error that the channel is undefined or if using the a defined channel from my config, I get "no code for module <module name>. Any ideas? <ieure>mdioc, "no code for module" means you're trying to use a module that's not on your Guile load-path. <ieure>mdioc, Are you using a variable from your channel to represent its channel record, and appending that to %default-channels? <mdioc>ieure, yes I figured that would be the case, but I can run guix repl and access those modules with no problem so I thought they would be in my load-path <mdioc>Correct to your question, that is exactly what I was trying to do to avoid duplicating the channel declaration <ieure>mdioc, Can you share your code? I'm wondering whether you're using a gexp to do the append; if so, the gexp is evaluated by the build daemon, and the load-path won't be the same as when you run guile (or `guix repl') locally. <mdioc>ieure, I am using a gexp to do the append <ieure>mdioc, Then yes, this is your exact problem. <mdioc>ieure, should this just be an expression to do the append, no need for a gexp? <ieure>mdioc, It depends what unattended-upgrades-service expects; I don't know what that is. But start there and see what happens. Or you can keep the gexp, but ungexp the channel you're appending. <mdioc>Cool, that is a good place for me to start, I'll give it a shot without the gexp and see what happens <ieure>mdioc, On whether this is the "right way" to do this: I don't think there is a "right way," so if it works for you, that's good enough. <mdioc>ieure, Fair enough. I came into guix complete scheme-ignorant and only ever poked around with nix a bit to understand the high-level concepts of declarative configuration, etc. so I am never really sure if I am creating something good or monstrous :) <ieure>mdioc, What I do is the other way around, I define a minimal operating-system for the specific machine (hostname, filesystem UUIDs, etc), then apply a series of transformer functions to it. <ieure>mdioc, So my config looks like: (define %os (operating-system ...)) (define xf (compose profiles/+exwm profiles/+bootstrap)) (xf %os) <ieure>More transforms than that, usually 8-10 for a typical system, but you get the idea. <ieure>What I like about this is that when stuff needs to touch multiple parts of the operating-system, all those changes are centralized in the one transform procedure, so it's very easy to add/remove stuff by adding/removing transforms. So you can have stuff like +intel-video, and that adds some packages, changes the kernel arguments, and pulls in some xorg configurations. <Noisytoot>ice-9/eval.scm:142:16: In procedure compile-top-call: <Noisytoot>(I can guix pull --roll-back and guix pull, but still) <Noisytoot>actually maybe if I upgrade guix now it'll be un-broken? I'll try <ieure>Noisytoot, What commit are you on? I pulled last night without issues. <Noisytoot>ieure: current system generation is 5863808183e11ae3f1ac1378dff8b079748ada3c <ieure>Noisytoot, "system generation" would `guix system describe'. I'm asking for the commit shown by `guix describe'. <Noisytoot>ieure: 28d6bcc930be7da72ee5e0a938a9a37c83ad2afc <Noisytoot>this is after I successfully pulled with guix pull --roll-back and guix pull, I haven't ran guix upgrade or guix system reconfigure yet <ieure>Noisytoot, I can't reproduce with `guix time-machine --commit=28d6bcc930be7da72ee5e0a938a9a37c83ad2afc -- pull --profile=/tmp/guix.repro'. Are you using third-party channels? They may be the source of your issue. <mdioc>ieure, your solution seems great, I may use it for 'inspiration' :) <ieure>mdioc, Thanks, please, rip it off! <untrusem>mine is working fine but i might use compose transformations more <gnoodles590>Does anyone know the proper way to run guix/hurd in qemu? <gnoodles590>I tried running qemu-system-x86_64 --machine q35 -nic user,model=virtio-net-pci -enable-kvm -m 2048 -device virtio-blk,drive=myhd -drive if=none,file=86pbr2w7zfzf31njwwihxs86adwbb6zv-hurd-barebones.qcow2,id=myhd, but I get stuck on sd0: Waiting to identify in the boot process <ieure>Noisytoot, I would audit both of those for getenv calls. <Noisytoot>neither of them call getenv, at least not directly <dariqq>Noisytoot: do you have getenv in your channels.scm? <Noisytoot>yes (I have (getenv "HOME")), that must be it <Noisytoot>what do I need to do to make getenv available there, to avoid having to hardcode the full path? <SaltyRust>anyone working on a neovim nightly package for version 0.13.0? <stephen0>What guix package will provide a tui/cli version of emacs? <hjolmir_the_peni>i see a message to guix-devel@gnu.org in my sent emails from 5 days ago, and i definitely have requested to subscribe, but don't see my sent mail in the archives and haven't heard anything about the subscription. am i just stuck in the moderation pipeline? <stephen0>I didn't even see that one in the list when I was looking for emacs-nox <ieure>hjolmir_the_peni, If it was your first email to the list, then yes. <mattl>stephen0: yeah the extra dash is a bit confusing, whereas it's called emacs-nox in Debian, etc. <dariqq>noe: Changing shepherd -0.10 rebuilds elogind <ieure>I agree issues is not responsive, but it's also not the current issue tracker. <Noisytoot>the bigger problem is substitutes being unavailable <Noisytoot>guix substitute: warning: ci.guix.gnu.org: connection failed: No route to host <attila_lendvai>the /gnu dir cannot be a symlink? how else can i link/move it to another disk where i have more space? <ieure>attila_lendvai, You have to mount it on /gnu. <attila_lendvai>i wish i could to avoid reorganizing the disks/mounts only for this. is this some security measure?