IRC channel logs
2025-11-21.log
back to list of logs
<hanker>Is there no `-E` for guix home containers? <lostInSauce>Hello I currently have enlightenment desktop using xserver and would like add niri, which uses wayland , I installed niri via guix packager than copied niri.desktop to /var/run/current-user/system/wayland-sessions/ but its read-only .. not really sure how to get niri to work or an option from GDM ? <lostInSauce>Hello I currently have enlightenment desktop using xserver and would like add niri, which uses wayland , I installed niri via guix packager than copied niri.desktop to /var/run/current-user/system/wayland-sessions/ but its read-only .. not really sure how to get niri to work or an option from GDM ? <avalenn>I have a "Page not found:/eval/latest/dashboard." message on pulls.ci.guix.gnu.org <Deltafire>lostInSauce: if you install niri to your system config instead of the guix package, then it will automatically show up in the gdm options <Deltafire>but note that niri is currently broken in guix master since yesterday <SuperGuest>it tells me that some variable are undefined and that I forgot to use some modules <SuperGuest>the issue is, those variables are defined and I use-module those module <SuperGuest>and it tells me about a variable and module that I dont even use in the config I am trying to use <Deltafire>(home packages) is not the same as (home-packages) <SuperGuest>yes ? I import is as (home packages) since the path is home/packages.scm and it exports home-packages (list of packages) <SuperGuest>why guile doesnt show from which file the error is from, I dont even know where to look at <SuperGuest>"error: home-packages: unbound variable" the only files I have used that variable do import (home packages) <avalenn>how can I obtain a list of packages from my profile (like with "guix package -I") with the channel from which they are originated <gabber>how can i share my screen through sway/wayland/firefox/bbb? i read online about some pipewire-media-session package, but this seems unavailable for guix <gabber>avalenn: i just stumbled over that thread! thanks! <aargoli>for some reason, building niri keeps failing on my machine. It fails on "install-completions" phase due to: <aargoli>starting phase `install-completions' <aargoli>In execvp of /gnu/store/4mndjfq1vbh2chclcgjgc399jkf4hg3g-niri-25.08/bin/niri: No such file or directory <aargoli>error: in phase 'install-completions': uncaught exception: <aargoli>wops, sorry, didn't mean to paste all that <aargoli>i don't know how to further debug this <aargoli>oh, I thought it might have been something local. I did manage to get it to build once, but then tried a system reconfigure and it kept failing at that step. Damn, ticket opened last night, I had searched for tickets before that ticket was up <Deltafire>aargoli: i tried to figure out what had changed to cause the build to fail <Deltafire>nothing in the package definition it seems, last commit was September <apteryx>The FSF40 Hackathon in starts in 3 hours :-) <aargoli>Deltafire: yes, I saw that too. What bothers me is that I did succeed to do a guix install niri last night, but the drv hash was different than the one I get today <aargoli>I rollbacked and accidentally overwrote that generation <aargoli>i am new to guix or nix, so idk why the hash would be different, since it was also the same version number <aargoli>i guess some rust dependency updated and broke it? <aargoli>ah, yesterday cargo-build-system was changed, so maybe that caused it. How can I set guix at a specific commit (before the changes) so I can try installing niri and see if it fixes it? <aargoli>tyvm! had to add --allow-downgrades too =) <aargoli>looks like that's it, I regressed guix to commit 82918e4642c61170094720785950a23ecfe90e30 and successfully built niri <aargoli>what would be the best path forward, make another issue calling out the regression that some commits caused? There were at least 3 commits that changed cargo-build-system yesterday. Or just leave the Niri ticket open with the comment about the regression? <Rutherther>creating multiple issues for the same base issue is leading cognitive overload, so I personally see no reason to make a new issue. Rather comment on the already existing issue <aargoli>thank you, that's what I thought as well, just that the title is a bit misleading. Deltafire could you maybe rename it? <aargoli>misleading wasn't the right word, I think we could rename it to point out that the changes to cargo-build-system is the cause <Rutherther>I see, yes, appending something like "after recent cargo build system changes" could make it get more attention from the right people <Rutherther>I don't think so as the init would do the same as reconfigure in this case, so it suffices <apteryx>it's surprising there is no emacs service for guix home <flurando>hi, I have problem setting up oci-service-type. <flurando>always no cgroup defined or no docker defined (supplymentary group) for user "oci-container" <flurando>must I configure rootless-podman service to use oci-service-type? <ptah>Is there an issue with the git repos? `guix pull` says 404 <flurando>well, I figured out that you must have docker+containerd+elogind running or equivalence in podman services running to use oci-service-type. <Rutherther>flurando: you need to have either docker or rootless-podman, docker being the default <apteryx>If someone is interested to join some live hacking (FSF40 hackathon) session starting in about 40 minutes, let me know! we'll experiment with scripting using the Guix API, packaging, attempt to reduce the closure of the bare-bones image template, and more. We'll be able to share ideas and help each other via a Galene instance (chat/video/screen sharing). <ptah>flurando: What command are you running? Running `guix pull --url=ssh://git.savannah.gnu.org/git/guix.git` with either `ed25519` or `RSA` fails. On the other hand, shouldn't HTTPS just work? I can't access Savannah even in a browser https://savannah.gnu.org/git/?group=guix <Rutherther>savannah really has issues for past few months, so it's better avoided <flurando>Onh, no. After deploying, I could not ssh to it anymore, any connection is closed upon created. I wonder if this is because using elogind on headless machine should be banned or my 2G ram is drained by docker and containerd <ptah>Rutherther: And use what instead? I haven't used `guix` in like a year :) <flurando>Now I have no way of accessing my server! Because ssh is the only way to connect to it. <ptah>Rutherther: Nah, I have a system running already, running a commit from Feb. 2025 `d7ca62b15de7ef89c88ef9b1118d29481ca50122` <ptah>Thanks, I will take a look at codeberg or the redirect. <Rutherther>ptah: then definitely the second link is the better one to use. And after you pull from it once with --url, subsequent pulls will default to it <ptah>Rutherther: Perfect, thanks <flurando>Found solution. Adding elogin requires a reboot or restart of service sshd. <flurando>Well, my low docker skill fault, not associated with guix <tusharhero>hello, can someone tell me how the hackathon is supposed to work? <pomel0_>I have a weird problem with pipewire, in that the progarms I install with guix are not recognizing its jack replacement, but the faltpak apps I install do <pomel0_>for example, trying out Ardour, the guix package only lists pulseaudio and ALSA as available Audio drivers, but the flatpak recognizes jack/pipewire too <robin>could the guix package be outdated compared to the flatpak version? <pomel0_>I also had this problem with reaper, which I installed through guix, although that version *is* outdated <pomel0_>both ardour in guix and in flatpak are version 8.12 <pomel0_>huh, weird. The Guix manual says `pw-jack` comes installed with pipewire, but I don't have that <pomel0_>aaand I can't install it either, it's not a package that exists <pomel0>the real problem is that I want to get reaper to work. I can use the ardour flatpak just fine, but there's no reaper flatpak so I have to use the guix package, and pipewire is not working there <kestrelwx>In a `guix shell pipewire`, I do see `pw-jack`. <kestrelwx>I want to give my home Pipewire `realtime` group, seems a bit involved. <pomel0>I didn't have pipewire installed. <pomel0>I thought having `(service home-pipewire-service-type)` on my guix home would install pipewire automatically <kestrelwx>Could probably write a service that launches `pw-jack` after the Pipewire service. <pomel0>so what does home-pipewire-service-type do that it doesn't install pipewire? <pomel0>but it must install something, since flatpaks were working with it <pomel0>or do flatpaks install a faktpak pipewire... <Rutherther>it gives you shepherd services that start pipewire, two other services and also adds a pulseaudio config file so that pulseaudio doesn't get started automatically before pipewire has chance to start <pomel0>Rutherther: ok, I understand that, but now I'm confused as to why, for example, qutebrowser had audio working. I'm guessing it was just using alsa or pulseaudio when it couldn't find pipewire? <kestrelwx>my.fsf.org is unreachable currently, isn't it? <Rutherther>pomel0: most probably it was using pulseaudio, yes. Pulseaudio server is started by the first client that tries to access it <pomel0>and why were flatpaks working and showing pipewire (even qpwgraph was working and showing connections), that also confuses me <aargoli>Rutherther: ty for the PRs to fix niri and cargo-build-system <kestrelwx>pomel0 Well, you have the server running, that's probably enough. <kestrelwx>Oh, hostux.social is throttled here apparently. <pomel0>kestrelwx: right, I guess flatpak installs its own pipewire. Weird thing was that qpwgraph, which I installed through guix showed the flatpak apps working just fine under pipewire, before I had pipewire installed lol <JodiJodington>tplaten: I looked at the website, what is it like? is it just a competition to contribute to some FSF projects of your choice at your own time? Or is it more structured than that <gargantua_kerr>I came here via fsf 40 hackathon but I see no list of tasks/not sure what to do. Could someone help? <neox>You can join specific IRC channels to know more about tasks ;) <JodiJodington>neox: it looks like for guix, the specific IRC channel is just #guix and it doesn't seem like there's any specific tasks listed here <neox>JodiJodington, you can just ask! <neox>Guix people will answer, don't worry ;) <tusharhero>can someone share the credentials for guix hackathon <tplaten>I am using (build-system cmake-build-system) in a package currently working on. How do I pass CMAKE_CXX_FLAGS? <tplaten>For today I am packaging a VR game called Overte, that I want to play on my Valve Index (using Guix) <gargantua_kerr>seems like git.savannah.gnu.org is down? `guix pull` is stuck as a result <Rutherther>but savannah is really struggling for past few months, it's better avoided anyway even if it's not completely down <Rutherther>it doesn't, guix hasn't been using savannah as primary repo for months, it's only a mirror <ieure>tplaten, It should be on your system already, you can use the terminal reader (`info'), or C-h i from inside Emacs. <tplaten>Where, I usually open it in the browser on gnu.org <Rutherther>it's running on guix infrastructure, not gnu, so it's still up <ieure>tplaten, If you're replying to me, your answer is in the message I replied to. <ieure>*in the message you replied to <FuncProgLinux>Another answer could be running "info guix" from the terminal emulator <FuncProgLinux>ieure: ah, just re-read that, it's true :/ but I would recommend on reading the manual on Emacs <FuncProgLinux>It's way nicer/friendlier/easier to navigate than the terminal emulator alone <ieure>FuncProgLinux, Yes, I agree, Emacs' info reader is very nice. But I'm a tremendous Emacs dork, and recognize that it's not everyone's preferred tool. <ieure>Though #guix probably has one of the most favorable Emacs-to-non-Emacs-user ratios of any channel on Libera other than #emacs. <FuncProgLinux>Funnily Guix forced me to not rely on syntax checkers and fancy IDE stuff when writing scheme lol <ieure>I think Guix heavily self-selects for people into Lisps, so, I'm not surprised by that at all. <tplaten>git grep xrgears does not find anything, so I package that one first <identity>FuncProgLinux: i think there is a few, at least, (neo)vim users in there, maybe kakoune, and plan9port is packaged so there is probably one or two sam/acme users <identity>a cult classic, still going strong from what i know <ieure>I gotta get my DECwriter III hooked up and learn ed(1). <Rutherther>so I have a store path where some files should be present, but aren't... but the hash in the db... matches with my contents in the store?! Seems the error must have happened during build or substitution. When I manually fetch the nar I see the missing files there. Really interesting <cbaines>given the missing files are in the nar, this suggests a build issue <Rutherther>it's libpoppler-glib lib files under /gnu/store/1vz4z72bx8wxrc757yzk3q6gismrknz7-poppler-22.09.0 <cbaines>it's possible there's an existing issue, but this could have also been introduced with the changes on python-team <Rutherther>it really sucks guix doesn't really have good means to recover from corruptions in general, but even moreso with corruptions where the hash matches what's in the db. Something I would like to make more useful tools for in the future. I went to the db, changed the hash to wrong one and ran guix build --repair to fix it :) Now I can build my icons because inkscape no longer crashes on missing library. <Rutherther>I now even saw in the db that there is information about the drv an output path came from, I didn't know that. So in many cases when users do have --keep-derivations (the default) it would be possible to rebuild a path when corrupt path has been found, but it's not really something guix supports <cbaines>the package built successfully, it just doesn't contain the files it's meant to <cbaines>Rutherther, it's possible you got that poppler missing files from bordeaux.guix.gnu.org, but only if you were using python-team before it was merged <Rutherther>like if I had the path in the store for longer time, yeah, that could be it <Rutherther>but it's a question whether it's really a problem in the build or if bordeaux suffered a corruption <Rutherther>or does bordeaux check the hash matches what it should with every download? <cbaines>unfortunately it's not as clear as it should be which nars where generated by which builds, but given there are build logs for the exact same derivation, with clear differences, it suggests that this is a problem in the build (or with an input, with some machine specific thing like the kernel version, ...) <ham5urg>any other source where I can get the ISO? I just made a ventoy boot stick ready. <ham5urg>git.savannah.gnu.org is down too. any news why this happens? <ieure>ham5urg, All of gnu.org is down. I don't know the reasons why this happens, probably too few people responsible for GNU infrastructure and not enough time to do it better. <ieure>gnu.org is down so hard, my email to help-guix bounced. <ieure>Two failed SSDs and a failed "front of rack switch." <ieure>Which I think is an Ethernet switch. <oliverD>is there anything to work on (I don't have much contributing experience so I'm not sure how this goes) <ieure>oliverD, Sure, you could review some PRs in Codeberg. Or look at open issues. <ieure>Reviewing PRs is probably the biggest impact for new contributors. <gargantua_kerr>Running `guix build -L gnu/packages hello` returns " Wrong number of arguments to" and errors out. Could someone help? This package is clearly present in gnu/packages/base.scm <ieure>gargantua_kerr, Your -L argument is incorrect. Also, -L doesn't work in the main Guix repo, for reasons I do not understand. <ieure>gargantua_kerr, `guix shell -m manifest.scm --pure', then `./bootstrap && ./configure && make -j8 &&', then `./pre-inst-env guix build hello' -- see the manual section "Running Guix Before It Is Installed." <abbe>./pre-inst-env guix build hello <ieure>gargantua_kerr, -L adds a path to Guile's %load-path. If you use (gnu packages base), it will look for a file named gnu/packages/base.scm relative to the root of each entry in that variable. Presumably you're running your `guix build' from the root of the main Guix repo, so that's going to add $PWD/gnu/packages to the load path. And there is no $PWD/gnu/packages/gnu/packages/base.scm <ieure>`-L.' is the correct invocation, other than, as I mentioned, it does not work. I wish it did. <gargantua_kerr>ieure: I did follow this and IIRC "./pre-inst-env guix build hello" did work. I then tried to change the name of the hello package to sth else so I can make sure its actually working. I changed it to helly. Now when I did `./pre-inst-env guix build helly` it failed. <ieure>gargantua_kerr, Please paste the error you're getting *into a pastebin*, not direct into the channel. <ieure>(Assuming it's more than, say, three lines)