IRC channel logs

2026-05-02.log

back to list of logs

<lullaby>My system gives a black screen sometimes on resume from sleep. How to debug? I've looked in /var/log/messages and can see a couple lines of gibberish ^@^@^@^@^@^ written right before I rebooted it due to black screen.
<Sven792>uuuh, that would be interesting for me too. I have a similiar problem on my laptop
<Sven792>I disabled auto suspend because it seems to be completely broken. It doesn't look like it will wake up properly at all lol
<Sven792>lullaby: can you see the kernel logs?
<lullaby>Sven792: Where are the kernel logs?
<lullaby>Maybe dmesg? Checking if dmesg has an option to view logs from previous boot.
<Sven792>journalctl has an option for previous boots. "journalctl --list-boots" and "sudo journalctl --boot=BOOTID"
<Sven792>oh, journalctl is part of systemd. that's not in guix
<Sven792>I tried "sudo dmesg -w > kernel-log.txt". I atleast don't get anything after the sleep
<bdunahu>if you have a spare machine, you could try to ssh into the one with a black screen?
<loquatdev>Hello, everyone. Why is it that using `(include filename)` in one of the modules inside my Guix configuration is invalid?
<loquatdev>I have a file containing channels (generated by `guix describe --channels=format`) that I'd like to include in a larger file containing other information such as signing keys.
<loquatdev>The problematic snippet is `(define %my-channels (include "channels-file.scm"))`. It works when I am inside the folder containing those modules, but when including it from another directory via -L /path/to/directory the module containing it ceases to work properly, giving me several `unbound variable` errors for things defined in that channel.
<loquatdev>By channel, I meant that module.
<lullaby>Sven792: Yeah journalctl was the first thing I reached for haha
<lullaby>bdunahu: ssh might be doable :)
<lullaby>Maybe I could `dmesg -W | tee some-file` and then after it freezes I could inspect that file
<lullaby>With `dmesg -W` running I've closed and reopened the lid several times. No issues yet
<sunless>hi, i'm working in a guix fhs container, and 'fusermount3' isn't working because of "operation not permitted", i've tried to change the permissions to the folder i'm targeting but it's still not working, how can i grant it permission to mount
<sunless>actually nevermind, the issue had nothing to do with the container, it's about fuse, kernel support and setuid requirements ... i'm changing my overall approach
<human_equivalent>Hello. I'm looking into an issue tagged "good first issue" about the guix pack command. I've found the source in guix/scripts/pack.scm, but I don't know how to properly test my changes. I've managed to change package definitions before but this is put together differently. Looking at the builds of guix on ci.guix.gnu.org doesn't reveal anything to me either. Can somebody point me in the right
<human_equivalent>direction?
<Rutherther>human_equivalent: see 22.4 Running Guix Before It Is Installed in the manual
<human_equivalent>Awesome, ty
<Yapppaholic>Hi everyone! I'm trying to get a patch for sched_ext merged into guix and wondering, how long can it possibly take for it to get approved and merged into upstream?
<ekaitz>Yapppaholic: it depends on how hard the patch is to understand and review by a commiter
<ekaitz>for example, if you send it to me, and i can review it fast enough, I could push it straight away
<Yapppaholic>ekaitz: It's a simple flag for kernel config and it's already pushed as a PR to guix, but cuirass bot ran out of memory while building the kernel so I couldn't get it approved (and I still need kernel-team approval). https://codeberg.org/guix/guix/pulls/8048
<ekaitz>Yapppaholic: in that case, it'll take long
<ekaitz>few weeks more
<ekaitz>people are busy
<Yapppaholic>ekaitz: I understand :(. Would be nice to get it merged in the future so I could extend it further with scx service and some rust-based schedulers
<Yapppaholic>Guess I'll have to wait then
<ekaitz>yeah... more than wait, you could send more patches of other things :)
<ekaitz>or start with your next ideas on top of this
<ekaitz>eventually you'll be a commiter and you'll be able to push everything you like
<Yapppaholic>I'm already running sched_ext on my system with shepherd service and scx_lavd (just that it takes ~2.5 hours to build kernel myself :D), currently I'm trying to fix native package manager for neovim with 0.12.1 release and wait for cuirass bot to evaluate changes
<Yapppaholic>Committer approval seems pretty hard for me, not sure if I would be able to get over 50 approved PR in guix
<nckx>I'll restart those failed Cuirass builds this evening.
<ekaitz>Yapppaholic: you don't need to have 50 approved prs
<ekaitz>that was an arbitrary number
<Yapppaholic>nckx: Thank you
<ekaitz>we need just to trust you and know you are a good actor
<Yapppaholic>ekaitz: Oh, I thought it was a hard requirement from the docs :D
<ekaitz>nah
<ekaitz>i don't know if I made 50 commits total and I'm a committer
<Yapppaholic>lol
<Yapppaholic>I guess I'll see how it goes, generally I'm not doing much kernel stuff, so getting into team-kernel is a no go for me
<nckx>If you ever change your mind, please do ;-) It's understaffed.
<Yapppaholic>Same with zig-team I heard
<ekaitz>Yapppaholic: feel free to join a team, it doesn't really give you privileges afaik
<ekaitz>you'll just get more email notifications :)
<ekaitz>that could let you be more informed or how things are done
<ekaitz>and if you can help with some specific task you could just raise your hand when the time comes
<nckx>Oh yeah it's not an access control method, it's a social construct.
<Yapppaholic>ekaitz: Do I need to create a PR with "/etc/teams.scm" changes? Also, do I have to discuss it with someone from the team directly or is it fine to go and send a PR without mentioning anyone?
<ekaitz>i would just do the pr, but I'm not very knowledgeable about the bureocracy
<nckx>Check out the manual section on Teams, but feel free to ask if something's still unclear.
<Yapppaholic>I see, I'll try to check if the wiki says anything about applying to the team/bureocracy things and then send a PR
<nckx>(Because that would be a bug.)
<nckx>Yapppaholic: There is no authoritative wiki.
<nckx>Which isn't to say the Libreplanet one is untrusted but everything important should be moved to the manual.
<Yapppaholic>nckx: Is https://guix.gnu.org/manual not an official source of documentation?
<Rutherther>Yapppaholic: ideally mark the team as reviewers and let someone from the team approve it
<Yapppaholic>I've used libreplanet only for substitute mirrors and nothing else
<Rutherther>Yapppaholic: the link you sent is official documentation. It's not a wiki though
<Rutherther>nckx: I managed to do you what you asked me to do :)
<Yapppaholic>Rutherther: I'm looking at one PR for team-emacs and it also requires team-core approval, is it necessary or (in this case) team-emacs approval would be enough?
<Rutherther>Yapppaholic: core team is just responsible for the file, not for members of the teams
<Yapppaholic>Rutherther: Ah, I see, thanks for the info
<attila_lendvai>if there's someone around who understands nested user namespaces, kernel security policy, seccomp, apparmor, etc... then here's a triaged recent breakage that could use that knowledge: https://codeberg.org/guix/guix/issues/7852#issuecomment-14178845
<attila_lendvai>the regression is that chromium based browsers inside flatpak fail without --no-sandbox. a reproducer is `flatpak run com.brave.Browser --no-sandbox`. with sandboxing enabled browser tabs die.
<sham1>Am I correct in my understanding that if I want to have a programmatically generated program file added to the store and the user's PATH I'd have to create a small package for it. Like I want to have a `program-file` (or `gexp->script` perhaps) output be something I can use as a program
<yarl>There is, I think, some inconsistency with the facts that 1/ the 'manifest' file is not in /var/guix/profiles/per-user/$USER/guix-home but under profile 2/ `profile-manifest` looks for `(string-append profile "/manifest")` 3/ `guix gc --list-roots`
<yarl>On 1/, all guix-home-*-link
<yarl>btw
<sham1>Alright, I managed to figure out what I'd have to do for a custom binary from a service. Taking a gander at gnu/services/xorg.scm, one indeed needs to have a small package with trivial-build-system and create an output dir in the store with a bin subdir and the program file symlinked there. For future reference
<achnazoor>hi.
<achnazoor>I've just managed to install guix os on my old yoga 370 (yay!), but I'm running into a couple issues...hoping someone here can point me in the right direction...
<achnazoor>1) Gnome - my mouse cursor appears to be corrupted: it defaults to a whitish rectangle...any ideas on how to fix this?
<achnazoor>2) Gnome + flatpak - installed flatpak using the official flatpak instructions for guix, but gnome isn't picking up on the installed stuff (found and implemented a fix described in a forum where I add the content of flatpak.sh to my bash_profile, but that tidn't help...)...any thoughts?
<achnazoor>Thx!
<ieure>achnazoor, I've noticed that Gnome cursor issue. It only happens on certain hardware. I haven't tried to figure out what's up with it, but it is annoying.
<mwette>I've been bitten by this cursor issue: https://wayland.freedesktop.org/libinput/doc/1.31.0/touchpad-jumping-cursors.html
<mwette>maybe it's related
<ieure>It is not related.
<achnazoor>so far as i can tell, the movement is smooth, it's just that whenever the thing is over a clear background, it becomes next to (or actually) invisible. I'd change the mouse cursor to a different image if I coud just figure out how to...
<yelninei>achnazoor: what is your XCURSOR_PATH
<yelninei>for flatpak are the flatpak direcotires in XDG_DATA_DIRS
<ieure>Is there a way to flatten the graph of a package's propagated-inputs? ex. the propagated-inputs of it and all its inputs? Figure there's got to be something like this in Guix somewhere already.
<Rutherther>ieure: that's transitive inputs, depending on what you're looking for the function will differ, the most generic one for all direct inputs (propagated, inputs, native-inputs) is package-transitive-inputs
<Rutherther>more commonly you might want only the target or native, though
<Rutherther>package-transitive-target-inputs / package-transitive-native-inputs
<ieure>Thanks.