IRC channel logs


back to list of logs

<podiki>ACTION is [m] user in all but name, hiding successfully
<dgr>got guix patched with and 4 and followed but still, if I do `sudo -E ./pre-inst-env guix system reconfigure ../../guix/config.scm` I get: "/home/dgr/git/guix/config.scm:91:10: error: (type luks-device-mapping-with-options #:/key-file.cpio): invalid field specifier" any ideas?
<apteryx>looks like you provided a record field that doesn't exist
<piptown>gdm doesn't seem to work with bspwm. I have chmod +x bspwm, sxhkdrc, and ~/.xsession with only exec. how could I get another login manager that will play nice with bspwm?
<piptown>rather: exec bspwm
<iyzsong>piptown: login as the normal user, did run "~/.xsession" leading bspwm complain about X server not available
<piptown>iyzsong: huh? or you saying to do that?
<iyzsong>yes, forget to add a ending '?'..
<iyzsong>~/.xsession need a shebang '#!/bin/sh' as its first line
<piptown>iyzsong: it says Another window manager is already running. Let me log and try logging in again.
<piptown>iyzsong: seems I'm missing something I got nothing but a black screen. polybar isn't  even displayed
<iyzsong>um, what's in your ~/.profile, does it source .xsession or start another wm?
<piptown>um, seems I don't have a ~/.profile but I have a ~/.bash_profile
<piptown>contents of .bash_profile:
<iyzsong>that looks fine, did you have a terminal installed (eg: xterm)? put 'exec xterm' in ~/.xsession to see if gdm can launch it to see the problem is in gdm or bspwm..
<apteryx>yay, yubikey-manager-qt is working
<piptown>iyzsong: yes, xterm startup just fine
<iyzsong>piptown: okay, so your bspwm may have some configuration problems
<viaken>What might cause a system reconfigure to hang just after "bootloader successfully installed"?
<viaken>I'm not doing anything fancy in the config.
<viaken>oh, it's herd I think
<ulfvonbelow>anyone know what the length limits are on user and group names?
<ulfvonbelow>I've discovered that when I try to run 'sudo -g' with a group name that's 24 characters long, it says I'm not allowed to execute whatever I was trying to run
<ulfvonbelow>this happens even if I first use sudo to become root
<ulfvonbelow>though it shows up in /etc/groups just fine
<ulfvonbelow>uhh never mind user error
<ulfvonbelow>forgot to change the group of the user in question
<ulfvonbelow>though I'm still puzzled why sudo refuses to allow me to mix and match users and groups
<Luk6655>hi, I'm trying to get guix to run on foreign distro(debian 12) in a docker container. It looked fine until I got: \builder for `/gnu/store/4xrbn6wh5m1rpj03zcxb224yzg7nf708-autoconf-2.69.tar.xz.drv' failed to produce output path `/gnu/store/n2d3rwbvrlag5c1spv1b73vvg9w2r8cw-autoconf-2.69.tar.xz during the first guix pull. Can anyone suggest some troubleshooting steps?
<Luk6655>nevermind, it looks like I needed to install netbase package first
<CompanyCTO>hi :)
<CompanyCTO>I've tried to install Jami (for very important CTO meetings), but it gave me this error: "GLib-GIO-ERROR **: 17:34:00.024: Settings schema 'org.gnome.desktop.interface' is not installed"
<CompanyCTO>I am not a very prominent linux user, so could you give me a hint about what package to install in addition to Jami for it to install this "Settings schema"?
<CompanyCTO>I remember that I had to install GTK+ when a similar error occured when using LibreOffice
<Luk6655>does anyone know what if guix has an equivalent of nix's sandbox = false ? My first guix pull fails with "cloning builder process: Operation not permitted" this is in a docker container btw, so I can't do "echo 1 > /proc/sys/kernel/unprivileged_userns_clone" (host is already set that way)
<hako>guix-daemon has a "--disable-chroot" option.
<Luk6655>is it the same think? I'll try it
<Luk6655>it seems to be taking a lot longer now (with --disable-chroot) so it looks promising
<iyzsong>CompanyCTO: install gsettings-desktop-schemas may help
<CompanyCTO>Thank you a lot! Seems like I won't get fired today
<CompanyCTO>(Should I file a bug report? Like, I think that Jami should have it as a dependency)
<iyzsong>CompanyCTO: yes, that would be great!
<CompanyCTO>I'll do this later this day then, after I figure out how to contribute
<iyzsong>sure, you're welcome :)
<hako>FORMAT: error with call: (format #f "missing modules for ~a:~{ ~a~}<===~%" tmpfs ===>#f )
<hako>🤔 "guix deploy" errors about missing modules when I set "none" or "tmpfs" for file-system-device.
<iyzsong>hako: yeah, seems likely a bug/bugs
<civodul>hako: hi! could you paste the complete output?
<civodul>it's trying to tell you the config is missing Linux kernel modules
<hako>civodul: I have pasted relevant file system and the output here: <>
<hako>The file system (/run) sets needed-for-boot? to #t, so it's checked by machine-check-initrd-modules.
<civodul>hako: thanks; to variable 'missing' is #f instead of a list
<civodul>so yes, setting (needed-for-boot? #t) gets it to be checked, and that check fails
<civodul>we can fix that.
<hako>(was afk for dinner) hmmmmm, handle them in (@ (gnu build linux-modules) device-module-aliases)?
<zamfofex>Does anyone know whether it’s possible to get ‘guix pack’ to generate a tarball without compression?
<zamfofex>It seems gzip takes quite a lot of the time while generating a large tarball!
<viaken>zamfofex: -C none
<zamfofex>Oh! Thank you! It just finished, but I’ll keep it in mind for next time.
<viaken>No worries!
<andreas-e>There is the "--compression=" option to "guix pack".
<andreas-e>The possible values are defined in gnu/compression.scm.
<zamfofex>andreas-e: Thank you! Seems like ‘-C’ and ‘--compression’ are aliases.
<andreas-e>gzip, lzip, bzip2, xz, zstd, none.
<andreas-e>Yes, they are the same. viaken beat me to looking things up.
<akonai>when sending a patch that adds a package and a dependency in the same file, is it required to split it into multiple commits?
<civodul>akonai: hi! yes please, that's what we usually do
<viaken>Ah, that's good to know. A commit per package?
<efraim>almost good news! I nearly got rust-bootstrap building on powerpc64le-linux
<apteryx>efraim: yay!
<apteryx>zamfofex: you could also use zstd; it compresses fast
<apteryx>what do I have to do for libusb to work without root privileges?
<rekado>apteryx: udev rule, maybe?
<apteryx>I think so! seems to be the 69-yubikey.rules rule provided by the yubikey-personalization package
<apteryx>I'll test
<XiaoBiHu>Hi, I'm new to guix and going through the packaging tutorial [1]. I think that the definition of `my-hello.scm` package should have an appended last line "my-hello" for the `guix package --install-from-file=my-hello.scm` to work correctly. Is that right?
<XiaoBiHu>If yes, I'd be happy to change it. Could you point me to some instructions how to submit the corrections?
<rekado>XiaoBiHu: no, the last value in that file is a <package> value
<rekado>when using “--install-from-file” that’s all Guix cares about
<rekado>whether that’s accomplished by evaluating a named variable or a plain <package> value doesn’t matter
<civodul>XiaoBiHu: also, check out a more up-to-date version at
<XiaoBiHu>when I run `guix package --install-from-file=my-hello.scm` i get:
<XiaoBiHu>guix package: error: cannot install non-package object: #<unspecified>
<apteryx>do I need to reboot for udev rules to become in effect? I thought that was fixed with moving them under /etc
<XiaoBiHu>unless I put the line there.
<apteryx>or maybe replugging the key makes sense
<apteryx>that was it
<apteryx>so adding the yubikey-personalization package to the udev rules of the udev-service-type service fixed it
<apteryx>(keepassxc can now detect by yubikey)
<jpoiret>XiaoBiHu: can you use to post your entire file?
<rekado>XiaoBiHu: sounds like you bind the package to a variable.
<rekado>“define” has an #<unspecified> return value
<jpoiret>XiaoBiHu: what rekado said: you need to add `my-hello` at the end of the file
<XiaoBiHu>yes, that's what I was missing in the tutorial and spend a considerable amount of time to figure that out. I was wondering if it should be added to the tutorial to make it clearer to other people.
<apteryx>efraim: are you going to merge the rust-team branch soon?
<apteryx>civodul: hello! I'd like to resolve #64746, but it's unclear which commit is supposed to work; also we've hit problems such as python2 from 2021 era no longer building (SSL certs expired); should be put the latest known good commit to a more recent date than that?
<apteryx>sadly we don't seem to have a binary substitute anymore for that old package, or if bordeaux has one it doesn't get used by the old Guix
<Luk6655>I've been trying to build an emacs-next relocatable package and my build process has been sitting on "building CA certificate bundle.." for last hour... (using 100% of only one CPU) is this normal or did something get stuck?
<Luk6655>cygwin is not "supported" beyond windows 7.... so much for cygwin
<Luk6655>I might try cross compiling it
<andreas-e>apteryx: What are your plans with respect to core-updates? CI seems to be relatively free, so it could make sense to build it out and merge it. In that case, I would suggest to include my patch from #64832.
<Luk6655>sorry, wrong windows
<andreas-e>Also there is a new gmp release, which would entail a complete bootstrap, I suppose.
<andreas-e>I am trying to persuade the mpfr developers to make a new release soonish, so that we might be able to combine the two updates.
<apteryx>andreas-e: I have docbook patches applied, wanted to put them to the test but I keep hitting a weird offload problem on my setup
<apteryx>(large llvm checkout transfer gets interrupted, with remote SSHD receiving some garbage)
<apteryx>andreas-e: feel free to merge core-updates things from the tracker to it
<apteryx>to push*
<apteryx>cbaines: according to our merge policy for branches, elogind-updates must be applied only after the rust-team branch gets merged, correct? And then after rust-team is merged, I should rebase elogind-updates on master?
<cbaines>apteryx, that's the current ordering since the issue for rust-team was opened first. You can discuss doing things differently with efraim though.
<cbaines>as for rebasing, you might not need to (if there's no interaction between the elogind-updates and rust-team changes), but the safe option is to rebase
<Luk6655>are guix builds single threaded?
<Luk6655>suppose I'm creating a guix vm for building software etc, is there any point giving it more than few cores? will guix build use them, or will they just sit idle?
<Luk6655>I'm not sure if this "single core thing" is a docker problem, or is it a guix problem?
<apteryx>rekado: I tried creating an account at; the javascript seems lighter, but still blocked by IceCat's LibreJS
<apteryx>at least it only requires an email address, which is better than
<rekado>apteryx: I didn’t notice any javascript during normal operations.
<rekado>I see that the login form has one trivial script:
<andreas-e>apteryx: I am pushing the wget patch to core-updates, since it has no reason to break anything.
<andreas-e>gmp should be tested locally first, although the changes do not seem too big.
<andreas-e>Luk6655: guix normally uses the available cores for building, unless this is disabled for a given package where parallel builds pose problems.
<andreas-e>apteryx: I had seen the ssh problems when offloading on berlin during the last core-updates merge, with big packages such as llvm, ghc or openjdk. I thought they had been solved in the meantime.
<andreas-e>After civodul did a wild strace session.
<viaken>I have a system where herd just hangs. I have a shell with `herd status` that's been stuck since last night. What info should I gather?
<Luk6655>is anyone aware of any issues running guix in a docker container? (other than it has to run without chroot)?
<viaken>apteryx: Could you share your udev service definition?
<Luk6655>It built emacs-next in 5 minutes, but then once I run guix pack, it has been taking a very long time. Last hour was spent on "building directory of Info manuals". We're nearing on 3h total...
<Luk6655>this is on a fast nvme drive, modern ryzen cpu and 32gb ram btw
<andreas-e>This is not normal, info and man pages should take on the order of seconds. Maybe stop it and start it again?
<andreas-e>But I have no ideas how Guix behaves in a docker container.
<civodul>Luk6655: compression with "guix pack" can take time, as has been discussed earlier today
<Luk6655>hmm, I'm worried it will take anothe 3h... I'll stop it in next few minutes probably, about an hour ago it progressed from making SSL certs to info
<civodul>3h seems like too much time though
<Luk6655>well, only last 1h was spent on info directory
<Luk6655>1h was building, 2nd was stuck on SSL certs, but it seemed to progress once I pressed enter so perhaps it was waiting for some entropy? the third hour is spent on info directory
<Luk6655>one cpu core is pegged at 100% so it doesn't appear stuck, just rather slow, this is in a docker container btw, with chroot off
<Luk6655>I'm planning to maybe leave it for few hours, if it doesn't finish I'll set up a guix VM and give it a go instead
<Luk6655>perhaps it is some issue between docker and guix?
<graywolf>Hi :) If you use home-environment to manage emacs config, how do you handle customizations?
<oleander>Hello! Does anyone get this error when trying to run wpa_cli: "Could not connect to wpa_supplicant: (nil) - re-trying"? I've tried to edit the wpa_supplicant.conf file adding the two lines suggested here: but didn't work.
<oleander>I can only run `wpa_supplicant -B -i interface -c /etc/wpa_supplicant/wpa_supplicant.conf` to establish a connection but not the wpa_cli.
<piptown>I think this is the line that maybe the problem with my bspwmrc file: sxhkd -c "$HOME"/.config/sxhkd/sxhkdrc &
<viaken>apteryx: Belay that! xD I got it. \o/ And thanks for the pointer!
<efraim>apteryx: I'm hoping to merge it as soon as says it's good. I have more changes queued up for the next round of the rust-team branch
<samplet>graywolf: I’m no Emacs expert, but no one is saying anything.... I just set ‘custom-file’ (in my Emacs config) to a file not managed by Guix Home.
<apteryx>cbaines: thanks
<apteryx>efraim: good, I'll just wait :-)
<graywolf>samplet: Me too, but it feels like a shame that the customized values are not versioned/synchronized across machines in the same way the init.el is. Not that I have any idea how to do it...
<graywolf>So I was curious if there is some clever hack I missed :)
<samplet>graywolf: Customizations are kind of a hodge-podge of configuration and cache. It makes it hard to manage, if you ask me.
<piptown>how do I make gdm let me switch between XFCE and bspwm?
<apteryx>rekado: yes there was some more javascript at login creation time, but nothing too bad
<zmhanham>Hello everyone! Was hoping on some guidance regarding guix home and profiles. I would like to use guix home to manage my home config. That is, my dot files, user services, packages. I'd like to split my desired packages into profiles that can be conditionally activated. Mainly, I would like to have a base profile with my common packages between my systems, a personal profile for packages on my personal machine, and a work profile with
<zmhanham>packages / services needed for my work. I would need to conditionally activate the work profile on my personal machine for whenever I need to work from home. For achieving this, the most straightforward way I can see is to create a base, personal, and work profile and write a shell script to activate them as needed. But I am curious if there is some way to integrate this with guix home. Any advice from anyone who has done similar
<apteryx>hello! the "shared connection" feature of NetworkManager seems to have been recently broken for me; anyone else using it?
<apteryx>the remote end stays "connecting"
<apteryx>i.e. not receiving a DHCP lease
<ternary>So I have a Rust program that compiles fine, but inexplicably cannot make a request to an SQL server when it's built by Guix. If it's built by cargo normally everything works fine
<ternary>Does anyone have any ideas on where I could start digging with this? I don't know how the guix packaging process could break that. Maybe some underlying network library?
<ternary>I don't get errors or anything, the program just hangs when it tries to make the request
<XiaoBiHu>I've just compiled a software package under guix. However, when I try to run it, I get an error that shared library was not found. Do you have any suggestions how to solve this issue, please?
<XiaoBiHu>I've just solved it by setting `export LD_LIBRARY_PATH=$LIBRARY_PATH`. I'm wondering why this variable was not defined and why there is non standard $LIBRARY_PATH variable in guix. What is the best way to overcome this issue?
<apteryx>XiaoBiHu: the guix way to fix this is to patch the dlopen or similar calls from the program; you can usually simply provide the absolute file name of the library instead of attempting to search it from the library path
<apteryx>you can grep the source for dlopen to see such patching in various packages
<apteryx>ACTION reboots again
<XiaoBiHu>apteryx: thanks!
<jpoiret>XiaoBiHu: in general, LD_LIBRARY_PATH and LIBRARY_PATH are two different standard environment variables
<apteryx>how are ethernet NICs supposed to behave? If I plug a network cable on one end; should some LED light up in this case? Something shown in dmesg?
<RavenJoad>Is anyone using cgit on Guix for web-browsing of Git repos? I am trying to configure cgit for myself and am struggling to get the behavior I want.
<viaken>zmhanham: I've not done so yet, but I figured I'd create records that have lists of packages, services, etc, then a mapping of hostname to setup, and finally a function to make the (home-environment) splicing it all together.
<apteryx>haha! my first Ethernet port (eth0) on my machine is named enp4s0 and the second one is named enp3s0. I always thought at least the kernel named things in the same sort order, seems I was mistaken there
<apteryx>so my NetworkManager error I was debugging for hours turned out to be PEBCAK; I was misconnecting things all that time ^^'
<bjc>did you swap out your cpu or mobo recently?
<apteryx>No, but I got rid of some devices sitting on my desk (and reorganized the wires a bit)
<bjc>sigh. if they're going to have horrible names, they should at least be stable as long as your hardware is
<bjc>as much as i don't like solaris' c0t0d0s0 nonsense, at least it worked as advertised
<apteryx>I'm still surprised by the eth0 -> enp4s0 and eth1 -> enp3s0 thing
<apteryx>I guess they are stable; I had both connected so when I kept just one I mistakenly thought it was enp3s0
<viaken>Theoretically the eth0 -> enp4s0 was supposed to make them more stable
<bjc>oh, i see. you went from 2 to 1, and bunged up which was which. i can't complain that much then
<apteryx>is there part of the Linux kernel doc explaining how it names network interfaces "stably" ?
<apteryx>apparently 4 in enp4s0 means it's in the 4th PCI slot
<apteryx>perhaps I should blame ASUS
<apteryx>or just paying more attention to the devices MAC addresses instead, eh :-)
<apteryx>apparently this naming is handled by systemd:; perhaps by elogind in our case?
<apteryx>or udev directly... not sure
<apteryx>yeah, it'd make more sense to be handled by udev
<piptown>finally got a setup that I can work with. Now I can enjoy guix