IRC channel logs

2025-09-06.log

back to list of logs

<c0co>does anyone know whether should we expect the xen package (as per virtualization.scm) to build the xen.efi and add it to the bootloader?
<FuncProgLinux>Rutherther: Will try that. I made a derivative package called "mate-extra" the one I shared yesterday :) thanks a lot!
<gabber>sneek: later tell c0co that easiest thing is to build the package and look at what outputs are generated
<sneek>Will do.
<gabber>sneek: thanks!
<gabber>sneek: botsnack
<sneek>:)
<gabber>do i understand correctly that etc/teams.scm is the single source of truth for which files belong to which team(s) and CODEOWNERS is in fact (automatically?) generated by that script?
<gabber>do i have to run `./etc/teams.scm codeowners` manually and add these changes to the repo?
<User-44>what's an easy way to run two instances of the same service type?
<gabber>User-44: usually just putting two of them in your system config file should suffice
<gabber>but you might need to tweak the configs of one of them so they don't interfere with each other
<User-44>says guix system: error: service 'hostapd' provided more than once
<User-44>Have different configs
<gabber>what exactly are you trying to do?
<User-44>run hostapd twice
<User-44>for each wifi antenna
<gabber>do you need to run hostapd for two antennas or do you need to run two different services for them?
<User-44>yes for different bands. 2.4 and 5. daemon for each
<gabber>it is possible that with the current way shepherd service are handled the latter is not possible (and if that is the case this could be a bug)
<gabber>can't it handle both bands with the same service?
<User-44>no
<gabber>then this may well be a bug (of our shepherd service)
<gabber>would you mind opening an issue on codeberg?
<gabber>and—if you can—craft a patch to resolve the issue?
<User-44>I don't think it is. I think services are supposed to be specifically coded to handle multiple configs if needed and hostapd-service-type doesn't. Just wondering if there was some work around. Thought simple-service might work but can't get it
<gabber>User-44: #guix is a bit lonely during night time. it is always a good idea to post your question to the help-guix mailing list—more often than not other people have run into the same or similar issues and know some tricks
<User-44>Thanks. Still day here, haha. Maybe someone is up and can chime in. If not I may reach out to the mailing list. Just seems like there'd be some way to just copy a service and use a different name with it so can spawn multiple daemons.
<gabber>ah, yes, sorry. euro-night (:
<gabber>i guess (but i'm not sure) that the symbol this service provides for shepherd here https://codeberg.org/guix/guix/src/commit/607dda2ccb097d0e2a6d04896c0f1534daab6216/gnu/services/networking.scm#L2402 has to be unique for shepherd to resolve its services as DAG
<gabber>it is absolutely possible to provide these symbols programmatically, i.e. as 'hostapd-wlan0 for the service on interface wlan0, then you could provide as many services (one for each interface)
<User-44>Thanks. I'm gonna dig into it. Never defined or extended a service before. See how it goes.
<gabber>it could be as easy as changing the line i referrenced above
<gabber>something like (provision '(hostapd-#$(hostapd-configuration-interface config))
<gabber>ACTION falls asleep—AFK
<apteryx>if someone is keen on troubleshooting a qtwebengine build problem, head to https://codeberg.org/guix/guix/pulls/1968
<apteryx>icecat and icedove 140.2 nearly ready, but help needed squashing a few problems: https://codeberg.org/guix/guix/pulls/2521
<c0co>We tried to brute-force multibooting with Xen. 1. Declaratively (multi-boot statement using old guix store paths for the xen and linux kernel images and initrd) and 2. (We know this is bad, it was desparation) going Root and manually editing the grub.cfg. Neither approach worked. We also read through the package definition for the current Xen package in virtualization.scm and couldn't figure out why it wouldn't generate the typically expected `xen.efi` binary into `/
<sneek>Welcome back c0co, you have 1 message!
<sneek>c0co, gabber says: that easiest thing is to build the package and look at what outputs are generated
<c0co>boot/efi`. I read through the latest CI/CD build log for `virtualization.csm` and the `xen.efi` does appear in the transaction several times, both being removed and then later handled on creation, so totally unclear if it should exist or not for Guix
<c0co>Now that I've had some sleep I was going to try to look through that build log more closely but seems like the site is having some teething issues atm. Lots of 502s (same with the XenProject right now too which makes this project tricky!)
<gabber>c0co: you can build the package locally and check the logs on your machine (they shall be equal)
<htgoebel>Hi, I'm late for switching to unprivileged deamon. Where can I find the instructions?
<identity>htgoebel: just set privileged? to #false in guix-configuration, which see: (info "(guix) Base Services")
<htgoebel>identity: I'm on a foreign distribution (sorry. missed mentioning)
<identity>htgoebel: then see (info "(guix) Build Environment Setup") § Daemon Running Without Privileges
<htgoebel>identity: thanks, this is what I was searching fot
<c0co>gabber: thanks, it's worth a try, but I am inexperienced with this. Where would I start to try to build the package locally, and guix-ly?
<Rutherther>thinking out loud here... should "(current-guix)" authenticate? The user is already using the given guix, so what's the point of authentication if they already agreed to use unauthenticated?
<Rutherther>(it currently authenticates and it makes it impossible for me to build the installer with my own commits as I am not a committer, I have to change the installer.scm source code to not authenticate)
<ArneBab>How can I efficiently test a servie in the guix repo? Means: without reconfiguring my system.
<Rutherther>in a container or a vm
<ArneBab>do you have an example execution for guix?
<Rutherther>no
<ArneBab>guix system container?
<Rutherther>yes
<ArneBab>Can I test a service as a userspace service instead?
<ArneBab>(I’m worried that I could break stuff when I run guix via ./pre-inst-env on my local checkout with my changes)
<ArneBab>… will experiment with container …
<ArneBab>Rutherther: thank you!
<abralek>Hi Guix!
<gabber>levenson: \o
<gabber>ArneBab: `guix home container` does the trick
<gabber>but i'm not sure if i understand correctly what you are trying to do since system and user services are not the same services
<ArneBab>gabber: I’m working on getting a service ready for a pull-request and need a simple way to test it. It’s a system service at the moment.
<ArneBab>Can I turn a system service into a home service without a lot of effort?
<serok>Hello guys how to fix this error: unrecognized boot parameters at '/var/guix/profiles/system-2-link/parameters' ?
<serok>guix pull doesn't help
<serok>this parameters file is empty, I think it could be something wrong with my btrfs filesystem because before I had some similar empty residuals but guix gc was cleaning it so I could move forward
<serok>besides that when I want to use vi editor I'm getting Segmentation fault, wtf?
<kestrelwx>Hi! I've just packaged 7-zip and I've been wondering whether I need to delete the unfree unRAR sources before building and whether I should list every license mentioned by the project in the package definition. I've done both so far, so should I send a patch at this point?
<Rutherther>kestrelwx: yes, all licenses go to license, yes, unfree sources have to be removed, probably with snippet in the origin. So if you've done both, you can submit a PR
<Deltafire>kestrelwx: isn't 7zip already packaged? or is 7z a different thing..
<Deltafire>p7zip
<kestrelwx>Deltafire: p7zip is a 3rd-party port and was last updated in 2016.
<Rutherther>why is check phase before install phase? Doesn't that unnecessarily increase the risk of having builds different for tests? #t vs #f? 🤔
<distopico>Hi, I have a questions about guix deploy, I have something like https://codeberg.org/guix/maintenance/src/commit/b62cc35409c73ae2d3498126a90e69854d25005b/hydra/thinkpad-deploy.scm#L1, where the os definition in a module and another for deploy, but I want to do `guix deploy my-deploy.scm` said "no code for module (os)" , with GUILE_LOAD_PATH works but is something that I need to do always?, is not guile detecting module in the same
<distopico>directory?
<distopico>I don't see references to GUILE_LOAD_PATH in the maintenance example they are adding it manually before run guix deploy?
<Rutherther>exactly, guile is not automatically adding modules in current directory. As for guix/maintenance, you can see here an example https://codeberg.org/guix/maintenance/src/commit/b62cc35409c73ae2d3498126a90e69854d25005b/hydra/deploy-node-129.scm#L29, they are executing with -L modules. That adds to guile load path (plus also to guix package path)
<distopico>Thank you Rutherther
<pizzapal>ok so going full dummy-noob-mode here: i have a package that i got from guix (swi-prolog, v9.3.8); i want to build a development version of the package from a certain vc branch/commit. what the heck do i do? what i have tried is pulling down the repository using git clone and copied the existing guix package definition to a file i named swi-prolog.scm and placed in the repo's root dir. i changed the relevant git-reference arguments
<pizzapal>in the package def and also ran `guix hash -rx .` with the repo root dir as CWD and replaced the sha256 hash in the package def with that. then i tried to install it with `guix package -f swi-prolog` but i get a hash mismatch error.
<sneek>pizzapal, you have 1 message!
<sneek>pizzapal, csantosb says: Sure, guix is a great package manager for all emacs things
<pizzapal>i'm pretty sure that i was going about this the wrong way to begin with, it was more of a wild swing
<c0co>so in a VM I'm having an issue after rebuild where for the first few moments after login, my system time is +1hr (and then quickly resets to be correct), which is causing issues (e.g. can't reboot)
<pizzapal>i need to define a package variant?
<Rutherther>pizzapal: so what is the current git-reference you have?
<Rutherther>pizzapal: if you want to use local source, you can just use "(source (local-file "." #:resursive? #t))" instead of origin with git-fetch method
<Rutherther>pizzapal: with the hash... I am not sure if guix hash can give you a hash for a git checkout tbh, you could try changing it to the one that it tells you is the correct one. But I am also thinking... depending on how you specified everything, you could be taking into account also the package definition itself, and then of course it would be impossible to match the hash as the hash changes when you change the hash in this scm file
<pizzapal>hm...
<vagrantc>where does "inherit" come from? trying to add --keep-derivations=yes and --keep-outputs=yes to guix daemon in my system config... and the example i used from another machine uses inherit ...
<vagrantc>and since both guile and inherit are plain english words ... makes it a bit difficult to search for