IRC channel logs

2026-09-07.log

back to list of logs

<apteryx>note to self; when debugging why the generic html updater can't find a release, instrument import-html-updatable-release
<apteryx>other note to self; this is a place where having logging would greatly help
<TheTaoOfSu>Does Guix have much impact on RAM usage? I'm considering using it for a server, but it's going to be pretty low resource, 1-2 GB
<apteryx>most intensive resources usage like cpu and ram is when running 'guix pull' (or building packages locally)
<apteryx>in normal system operation it stays out of the way; with perhaps its pid 1 using some extra 30 MiB for the guile VM than say, systemd
<apteryx>(assuming you meant running Guix System)
<apteryx>I use it on a 2 GB vps without any problem; when you use 'guix deploy' it'll build most things locally, so if your local machine has better specs, it'll be fine.
<apteryx>TheTaoOfSu: ^
<TheTaoOfSu>Nice, thanks, apteryx
<coopi>meow
<coopi>gcd001 tells to announce new submitted gcds
<coopi>in info-guix
<coopi>but
<coopi>coopi does not have the ability to post there
<coopi>(its mail gets rejected since its no maintainer)
<coopi>how do i proceed??
<apteryx>coopi: you can use guix-devel
<coopi>apteryx: hmmmm, coopi had already announced on guix-devel
<coopi>does that mean it don't have to do anything further??
<tusharhero-xmpp>coopi: BTW what is your gcd about?
<apteryx>I see; I acked it in mailman now
<coopi>tusharhero-xmpp: https://codeberg.org/guix/guix-consensus-documents/pulls/15
<coopi>its been a couple days and no one has replied or even reacted
<coopi>so im worried that i did something wrong
<coopi>am i just being paranoid??
<apteryx>the info-guix mailing list is very low volume, and most things that is moderated (since nobody ever rarely posts to it)
<apteryx>so it needs an attentive human in the loop
<jlicht>coopi: just because you're paranoid, doesn't mean they're not out to get you ;)
<janneke>byte-code: Loading file /home/janneke/.config/emacs/lisp/guix.el failed to provide feature ‘guix’
<janneke>oh wait, that's a file by me -- hmm
<janneke>what happened to emacs-guix, is the question
<janneke>so, this (require 'emacs-guix nil t) needs to be changed to
<janneke>(require 'guix nil t) this now
<apteryx>hm, rebooted into GNOME 50, but GDM appears to auto logs with gdm-greeter, as if it didn't know about other users.
<coopi>jlicht: by 'did something wrong' i meant that i messed up the process somehow owo
<coopi>cuz
<coopi>ive already managed to do that once
<coopi>its a somewhat long story but due to a comedy of errors i had managed to exhaust six of my 7-day find-a-sponsor deadline cuz i did the announcement process incorrectly / in the wrong order
<user220>hello! does anyone know how to set up a different shell that is not bash or zsh? i have issues with env variables...
<joostvb>user220: what's the issue? i'm curious
<joostvb>(and i know about shells)
<user220>@joostvb env variables are not set...
<joostvb>maybe "export" them?
<joostvb>user220: ^
<coopi>user220: what shell are you using??
<coopi>for instance
<coopi>i use "schemesh"
<coopi>which does not source /etc/profile and hence does not set up the environment variables
<coopi>so i work around it
<coopi>by spawning the POSIX sh to capture and return the environment which i then import into the main shell
<coopi> https://codeberg.org/coopi/velcelis/src/branch/main/src/velcelis/home/services/schemesh.scm#L148-L175
<untrusem>ohh coopi in irc
<untrusem>did you find sponsors for your gcd? I can do that if not.
<untrusem>ACTION is very sick currently :(
<coopi>untrusem: owo
<coopi>haii
<coopi>alsu no worries about the gcd, i found sponsor
<coopi>hope u get well fast meaow :c
<user220`>@coopi thx. is there no neater way?
<coopi>user220`: if there were a neater way, i wouldve been using it instead :p
<coopi>this workaround is so far the most elegant solution ive found
<coopi>and it works perfectly fine for me
<user220`>m
<user220`>@coopi just to clarifi when i say shell i mean what shows up right after login...
<empathy_monster>Just checked the irc logs. Thanks so much cellarweasel for the info for the guix-install.sh script. Works like a charm!
<empathy_monster> It's been quite a while since I've run guix and I'm going through the docs rn. Curious if there's a way to distinguish between a native binary and one from guix? Bc atm my system seems to favor anything installed by 'guix install' and native binaries, say in /usr/bin/, take a backseat
<bremner>Is the need for a local keyring branch documented somewhere? I saw some mailing list discussion from 2020 but I don't really understand what the issue is.
<coopi>user220`: yes, me too?
<coopi>im referring to the login shell
<coopi>my workaround applies to that
<coopi>theres also an alternative approach, however, where you don't set your desired shell as the login shell at all. and instead, in your bashrc u configure the login shell to immediately spawn your desired shell, thereby inheriting the environment
<sham1>empathy_monster: you could just put your guix PATH etc after your system PATH
<sham1>You really can't pick and choose
<bremner>I guess a less XY question is "how do I run guix system image" with one new commit?
<bremner>I am getting stuck on authentication issues. I got as far as "guix git authenticate" passing, but guix system image is still failing https://paste.debian.net/hidden/1e923ab2
<bremner>I guess I need to define a channel?
<sham1>I guess you could go and add yourself to the keyring branch locally. Then it should work (assuming you haven't done so already)
<sham1>And/or have you signed the commit at the HEAD
<bremner>yes to both
<bremner>although maybe I did the keying wrong
<bremner>I just dropped a file on the branch
<PotentialUser-91>I ran into a kernel regression on linux 7.1.10. I had 7.1.5 before and no problems, now I get nvme timeouts and the system finally hangs. Old kernel generation works though. config.scm is no difference. Anyone else experiencing this?
<empathy_monster>sham1: Thanks, that seems to be a decent compromise.
<empathy_monster>I'm still working through the docs. I plan on installing the full system when I have some extra time to devote to it. Rn I'm just running guix pm on top of Arch
<empathy_monster>I'm mildly allergic to s-expressions :P so I want to spend more time with guile
<mfg>How do I make certain packages extend env vars like C_INCLUDE_PATH?
<mfg>Does someone have an example package I can look at on the top of their head?
<mfg>Also, how do I add a team? Can non-committers be team members?
<cbaines>mfg, search paths like C_INCLUDE_PATH are based on where files appear in package outputs https://guix.gnu.org/manual/1.5.0/en/html_node/Search-Paths.html
<cbaines>mfg, as for teams, yes, non-committers can be team members
<mfg>cbaines: thanks, I'll read that :)
<apteryx>mfg: non-committers can be part of teams yes
<apteryx>to add a team, you modify etc/teams.scm and you open a PR
<apteryx>or you discuss it on guix-devel to see if others would like to join first
<mfg>apteryx: thanks, for the suggestions
<barrbaa`>still dreaming on hurd
<bremner>err. now It seems to be building. Maybe it was as silly as being on the wrong branch
<bremner>beginner mind sometimes means I don't know when to expect success.
<bremner>guix system: error: /home/bremner/software/guix/gnu/packages/web.scm:6184:2: package `jq@1.8.2' has an invalid input: ("_" #<syntax-transformer oniguruma>)
<user220>@coopi why not source etc/profile instead?
<joostvb>bremner: at least the error messages are interesting.... (scnr)
<graywolf>Does --keep-going not work together with -P or am I just holding `guix build' wrong?
<ieure>graywolf, I think it works, maybe the rest of the packages you have to build can't be built because one of their inputs failed?
<graywolf>Maybe? Though every time I re-run it, it at least downloads some additional substitutes, so it seems there is more work that *can* be done. Hm.
<graywolf>Now that I think about it, it would be useful if -P had some way to exclude known-failing packages
<jlicht>efraim: uvwasi also fails for me on i686-linux, yet the test seems to succeed when not running in the container. Is there some way to get a coredump from withint the guix build process?
<jlicht>efraim: ah found it with good ol' objdump; seems the tests were only succeeding on x86_64 due to luck; relevant stuff was in a assert(..), which does not play nicely with our NDEBUG
<untrusem>jlicht, your quickjs build system pr is super cool
<jlicht>untrusem: thanks, I spent *quite* some time on getting the llhttp "replacement" in a workable state. Just happy it bore fruit
<untrusem>:D
<bascht>For the record: Asked a few weeks ago what the best way to start a graphical service that needs WAYLAND_DESKTOP with `herd`, if the window manager itself is started by the same shepherd session and https://codeberg.org/hako/Rosenthal/src/branch/trunk/modules/rosenthal/home/services/desktop.scm#L74 seems to be the answer to my question. :D
<Elwood_Soup>Hi y'all! Looking for some help installing Guix on a RockPro64. I have Debian running on the SD card and I'm trying to run 'guix system init' to install it to the EMMC. Getting an error about an unbound variable even though I've added the appropriate module under use-service-modules. Any help would be appreciated. SCM and log to follow.
<Elwood_Soup> https://privatebin.net/?4bd6b472b960538f#CQivioEZ3LDmjc9eeYczx9tnf41vCpJsrsMpw48gdsFU
<Elwood_Soup> https://privatebin.net/?ce20ab00f19f6258#9QxmCUrqzqfnC4oszBumj4pZewv2jtYsWMUuF3UVETEY
<ieure>Elwood_Soup, Your config seems fine, (gnu services desktop) is where plasma-desktop-service-type is defined. What commit of Guix does `guix describe' show? I vaguely remember the KDE stuff being in a different module at one point, maybe you're on an older commit and it's not where you expect?
<Elwood_Soup>ieure, I ran 'guix pull --url=https://codeberg.org/guix/guix --branch=version-1.5.0' so that should be where I am
<Elwood_Soup>For context, I'm assuming the problem is related to me running guix under sudo, but running it without sudo gave me an 'i/o file protection' error
<Elwood_Soup>i/o error on bootloader install after 'guix system init' https://privatebin.net/?5393b117370b820b#9UMjz4wYnRTnhnH3oA79ece6dByjaCTC8ns69wvxTo5r
<Noisytoot>Elwood_Soup: can your user write to /mnt/emmc?
<Elwood_Soup>Noisytoot: yes, /mnt/emmc is owned by my user with 777. It copies to /mnt/emmc for quite a while before it fails. I think it fails trying to install the bootloader.
<barrbaa`>.