IRC channel logs
2024-11-14.log
back to list of logs
<meaty>I got gzdoom packaging almost done, the last thing remaining is that the program still looks for files in /usr/share; specifically it auto-populates its config file with that in its search paths. Thus, it can be worked around by just editing the config file after installation, but how can I patch the package so that it looks in the /gnu/store/...gzdoom/share/ instead? <meaty>/gnu/store directory names are constant across systems, right? So I can just exchange one hard-coded string for a compile-time one and substitutes will still work, but any tips on doing this? I am unfamiliar with the gexp system <meaty>in fact, it seems all I need to do is redefine a macro in the source code, what's the most kosher way to do it? A patch for a single line seems like overkill <vagrantc>for simple one line patches you can use substitute ... <vagrantc>meaty: isc-dhcp seems to have an example kind of similar ... search for update-config-scripts-for-bind for an example... <vagrantc>oops, i mean the post-configure phase... <vagrantc>but it shows how to change a string in a file into something that includes the eventual output path <vagrantc>might be better examples, but that was the best i came up with with a little 'git grep -C5 substitute gnu/packages/' <vagrantc>meaty: there are quite a few examples in gnu/packages/games.scm ... just search for /usr/local <vagrantc>that is mostly how i figure out how to do this stuff ... guile/scheme/lisp not really my thing, but there are lots of examples to look at for ideas <sturm>survey done! thanks for organising this! <PotentialUser-34>hello, my yubikey doesn't work (particularly gpg). I think the problem is this line in the pcscd logs: version `GLIBC_2.38' not found (required by /var/lib/pcsc/<snip>/libccid.so)`. I see glibc was upgraded a couple months ago in 70b2015ec5. I think I could substitute the dependency and push it into the service declaration of pcscd, but poking <PotentialUser-34>around in the sources of ccid I see no hard references to glibc 2.38 (nor did I expect one). What's happening here? My best guess is that the binary wasn't rebuilt as part of the upgrade, but that seems kinda.. unlikely. <PotentialUser-34>a quick test of the theory I've got is to force build libccid and refuse substitutes for it, right? <PotentialUser-34>yep I bumped ccid from 1.5.2 to 1.5.3 -- no error log on restart of pcscd and reentry of yubikey, gpg --card-status works, I can decrypt things now. <PotentialUser-34>knowing what I know about guix I don't know why it was searching for an old version of glibc <Ironsmith>hi guix, how do i enable bluetooth? i added service (service bluetooth-service-type), reconfigured my system, rebooted, and i see service started with `sudo herd status bluetooth` <Ironsmith>issue is, when i try to toggle bluetooth to the 'on' position in the top-right gnome dropdown or the gnome setting it stays in the 'off' position <Ironsmith>do i need to part of any specific group? some people say i need to be part of 'lp' but what is this group? couldn't find info in the documentation <wakyct>Ironsmith, my BT is (service bluetooth-service-type (bluetooth-configuration (auto-enable? #t))) and I am part of the lp group <wakyct>I also explicitly add blueman and bluez as packages though I'm not sure if that's required <wakyct>when I login I always get an error related to BT but it works and I haven't bothered to investigate <Ironsmith>do you know what that group is for? i checked online and for ubuntu and arch it's for printer access <wakyct>according to that issue it said for d-bus access <wakyct>I'm not sure what that means tbh ;) <wakyct>I'm guessing lots of services use d-bus <wdkrnls>Dear Guix, can someone tell me if there is a modern version of mbox? I saw bubblewrap in the guix repo, but it seems harder to use out of the box. <wdkrnls>Obviously, there is also guix shell -C <wdkrnls>But the feature which looks most appealing to me about mbox is the interactive transaction mechanism <PotentialUser-56>hi yall, sorry for the guest account - Ill work that stuff out later, for now im just trying to unstuck a system. <PotentialUser-56>for background, I'm running debian bookworm, and using guix home to manage my dotfiles and dev dependencies.This is a brand new install, and I have done very little to it. <PotentialUser-56>1. I never know which profile I'm in - theres ~/.guix-profile, which points to /var/guix/per-user/foo/guix-profile, ~/.config/guix which points to /var/guix/per-user/foo/current, and ~/.guix-home, which points somewhere in the /gnu/store. should there be one I just pick and make canonical? (preferably guix-home). <PotentialUser-56>2. I don't know how to properly activate my environment. I try to `guix home reconfigure` and only get `module not found` errors now. Ive tried running ~/.guix-home/activate, sourcing ~/.guix-home/setup-environment, setting my GUIX_PROFILE to ~/.guix-profile/ and sourcing the profile there, and still the same error. I have no idea what I'm doing <PotentialUser-56>3. my gnome installation (which i installed using apt) is borked after updating some guix packages and/or installing `vim-full` from guix. I have no idea what the issue is but whenever i login the gdm session immediately closes with "gkr-pam: unable to locate daemon control file". I'm guessing some environment variable is wrong. <KE0VVT>Installing Guix on Fedora Silverblue. :D <jaft_r>Rutherther: you possibly around (and free to help, of course)? You tend to know this stuff in far more depth (plus I haven't ever used Guix on a foreign distro so I fear I'd definitely get things wrong); PotentialUser-56 could use some guidance. <PotentialUser-56>(no rush, I appreciate the help) this is my first time kinda committing to a system around guix. <PotentialUser-56>wdkrnls my home config works (or did work), but now for wtvr reason the load paths are broken and it wont run <PotentialUser-56>literally just installed `vim-full` (from `vim`), ran a `guix-pull`, and a `guix package -u` and now i can only login on a tty <jaft_r>Does removing vim-full fix anything? <wdkrnls>and you cannot do guix package --roll-back? <PotentialUser-56>jaft_r - i wanted to do that, but i can't just remove it from my home-configuration.scm and reconfigure, because it errors with 'module not found'. sourcing `setup-environment` or running `activate` also doesn't work, because `'vim-full' not found in profile. <wdkrnls>It's bizarre to me you need to reboot :/ <wdkrnls>FYI: my .config/guix/current shows current -> /var/guix/profiles/per-user/$USER/current-guix <wdkrnls>Honestly, I think my profile is kind of busted too since I see two different profile objects as well. <wdkrnls>I've noticed I have some trouble installing some softwares. <wdkrnls>However, usually I just see a guile error. <wdkrnls>And the stuff I care about still works :/ <PotentialUser-56>wdkrnls i did 3 rollbacks, set my profile to ~/.guix-home/profile, sourced it, and `"vim-full' not found in profile` <PotentialUser-56>and running activate and setup-environment dont fix it to let me remove it <wdkrnls>I haven't sourced a guix profile in a while. I tried taking that approach in the past based on Ambrevar's guide in the Guix Cookbook, but that wasn't reliable so I stopped. <wakyct>the only time I source a profile is after some guix pulls/upgrades I think? When the help text suggests it. Otherwise I just keep everything either in system or home <wakyct>I think it's done in your shell rc and profile files <wakyct>(I know all these diffferent uses of the term profile and user gets confusing...for me it does at least) <wakyct>when you set up Guix home IIRC it adds this stuff to your shell startup file <wakyct>btw, there's nothing wrong with sourcing profiles...I think some people do this or have done it. Personally I don't need that fine-grained separation <wakyct>there might have been more people sourcing profiles before the new developments in 'guix shell'? I dunno <wakyct>I think it gets a little confusing because some stuff in like the Cookbook and older configs might not be how stuff can be done now <PotentialUser-56>for some reason it finds the modules now (no idea), so i removed vim-full, reconfigured, rebooted, and gnome is still borked <PotentialUser-56>`"gkr-pam: unable to locate daemon control file` or similar (im typing on diff computer) <wakyct>oh you're on Bookworm...sorry I haven't done foreign distro <PotentialUser-56>i tried unsetting stuff or manually setting them in /etc/profile.d/guix.sh but nothing <wakyct>funny enough I was contemplating switching to Debian to make installing some software easier ;) <PotentialUser-56>still no idea why i can now run `guix home` again. ill leave that problem for later - but does anyone have a guide or link or something for debugging guix/guile loadpath stuff? <PotentialUser-56>ive also tried running dpkg-reconfigure, reinstalling it, as well as removing gnome entirely, purging all its dependencies, and reinstalling <PotentialUser-56>hmm, this forum topic seems to say its not an issue - in my case its the only `p3` journal msg and happens at session initialization <mange>Sorry to chime in late, but I have this in my "Guix on a foreign distro" configuration: if [ -z "$XDG_DATA_DIRS" ]; then export XDG_DATA_DIRS="/usr/local/share/:/usr/share/"; fi, which I think I originally needed to make Gnome work on Debian. <mange>That's in my .profile file, before it sources any Guix profiles. <PotentialUser-56>.profile is read-only, do you add as a dotfile to guix-home or something? or use the system-wide /etc/profile.d/guix.sh ? <apteryx>is 'match-record' supposed to work against plain srfi 9 records? <apteryx>I'm getting: guix/scripts/refresh.scm:380:16: error: version: invalid field specifier <mange>Ah, when I set this up I wasn't using Guix Home. I think adding it to the system-wide config would make sense. The value it's setting for XDG_DATA_DIRS is the documented "default" value to use, so it should be safe to do system-wide. <wakyct>yeah, I also setup desktop stuff in system. <PotentialUser-56>`XDG_DATA_DIRS=/home/$USER/.guix-home/profile/share:/home/$USER/.guix-profile/share:/usr/share/gnome:/usr/local/share/:/usr/share` <apteryx>nckx: I'm working on bettering --target-version, so that it accepts partial versions <PotentialUser-56>i changed it system wide, but it doesnt actually set it, im guessing guix home is overriding it somewhere else <PotentialUser-56>im guessing `setup-environment`, which is read-only, but even there the clause doesnt look like it should match <mange>Are you using GDM on wayland? I vaguely remember that there's something weird about how it sources profiles (maybe it doesn't?) for the user session. <PotentialUser-56>no wait, that wouldnt be it (because XDG_DATA_DIRS still includes my guix-profile even when i log in on a tty <Ironsmith>ah, i now know why my bluetooth dongle isn't working, there is only proprietary firmware for it: `Bluetooth: hci0: RTL: firmware file /*(DEBLOBBED)*/ not found` (as desired) <Ironsmith>any good suggestions for a usb bluetooth dongle that is compatible with linux-libre (or how to find such a dongle)? <unwox>Ironsmith: dongles based on CSR8510 chip should work. there are plenty of them on aliexpress. i've got one of those to use with my bluetooth keyboard and it works fine <divya>Also Ironsmith can I know which firmware is your dongle using? I'm in the process of reverse engineering some Realtek chips. <Ironsmith>sure divya! when i lsusb it lists my device as: `Bus 001 Device 010: ID 0bda:a729 Realtek Semiconductor Corp. Bluetooth 5.3 Radio`, according to the store page it's listed as using RTL8761BW <divya>Ironsmith: Thanks. I think this will cover the chips I'm aiming for. <futurile>divya: have you considered trying your printer out with another Linux distribution, either in a VM or booting a usb <futurile>divya: seems like that's one way to see if 'someone else' has a solution <divya>futurile: I've used the same printer on Arch for years. <divya>It used a hplip-plugin, probably proprietary or something. I tried the nonguix channel for the same, but doesn't seem to work. <futurile>divya: I was wondering if it was on the printer end <divya>futurile: If I give it a printing order by flapping its cover a few times, it prints fine :) <futurile>divya: hah hah hah - printers are definitely "magic" - I assume there's some printer magician chained to a desk somewhere in HP and that's the only one <divya>futurile: Printers used to be my favorite thing as a person who prints and binds his own books, but these corporations have made it terrible. <futurile>divya: oh wow, printing/binding your own books sounds very cool <nckx>janneke: [At least] your guix-patches subscription has been disabled due to excessive bounces from your gnu.org address. I don't have more information; the server just says '550 5.7.1 Mail rejected.' Usually a policy issue. <janneke>my email situation has been terrible the past months <PotentialUser-16>I am not able to test it (have not yet figured out how to try patches), but if someone could take a look at it and merge it if it looks fine, I would be really grateful! <janneke>ACTION prepares a gdb-15.2 packages with support for the 64bit hurd <orahcio>Hi, thank you very much nckx and Rutherther, I could to use the updated guix after set $HOME/.config/guix/current to my PATH variable. I choose guix on hard mode, inside nixos and fish :) <Rutherther>orahcio you should rather "GUIX_PROFILE=$HOME/.config/guix/72943 current; source $HOME/.config/guix/current/etc/profile" to get everything the profile has, not just its PATH <nutcase>I have home-mcron-service-type in rather minimal my home-environment. If I log in as the user, log out and log in again, I have more mcron instances (two) than I should have. The two instances of course fight for locking files, because both are trying to execute the same job definitions. Why aren't they properly shutdown, after I logout from shell? And why is another instance started, when I log in, although there is already an instance running?? <orahcio>Nice Rutherther, I'll try it, thanks <stochastic>> guix pull: error: Git error: unexpected http status code: 502 <look>stochastic: no, I'm getting these too, looks like somethings wrong with savannah <avalenn>is there any recipe to build some custom ca-certificate-bundle based on nss-certs but with some certificates added ? <peterpolidoro>I have started getting an error message that I have never seen before: "guix time-machine: error: execl: No such file or directory" <avalenn>peterpolidoro: without more information it is a bit difficult to answer, it seems some executable file is missing <peterpolidoro>I am just trying to guix time machine with a channels file to create a guix shell with a manifest file. Both files exists and it has worked a thousand times, but all of a sudden it is giving me that error <apteryx>avalenn: aren't certs just a collection of files? <avalenn>apteryx: quite but not only, each cert is in a file, but in order to be usable by (quite) all software the content of $GUIX_PROFILE/etc/ssl/certs must have : <avalenn>one file per certificate with a human-readable name (e.g. HARICA_TLS_ECC_Root_CA_2021.pem) <avalenn>one file per certificate with a hash as filename (e.g. ecccd8db.0) <avalenn>and one bundled file with the concatenation of all certificates (ca-certificates.crt) <avalenn>If I want to add one certificate, I think I can easily add the two single-certs files but have no idea how the bundle is generated <avalenn>And the bundle is what I need to be exposed via env vars CURL_CA_BUNDLE, SSL_CERT_FILE and other related <sleepydog>look for ca-certificate-bundle is rebuilt automatically when your profile is changed. you might not have to do anything <sleepydog>if your package puts certs in /etc/ssl/certs the way you describe, the hook should find it when you install the package <sleepydog>(i'm just putting this together reading guix/profiles.scm, someone here might know better) <avalenn>sleepydog: thanks, I will just try that and tell you back <clarkf>is there an emacs home service? something like a home shepherd service that runs 'emacs -d' on login? <Rutherther>clarkf: there is no emacs home service in guix channel. I think there is one in rde channel <clarkf>it doesn't seem worth subscribing to the rde channel for just that, I reckon <Rutherther>clarkf also a note - a service would not do emacs -d, it would do emacs --fg-daemon. It needs to know the pid of Emacs process and the easiest way to obtain it is to not let the external command spawn it <dthompson>does anyone know what to do when you wrote a buggy service and now shepherd won't respond to commands :) <dthompson>I have a service that failed to spawn properly and now everything is locked up <Rutherther>dthompson maybe there is a better way, but one would be to boot to older generation <Rutherther>PotentialUser-42 if you mean you downloaded the ISO from that page, then that one is for installation, so it doesnt use any external fs. You would need to install from it to your virtual disk first, and boot from that disk <clarkf>I've never used the official qemu image, but the name of the devices varies depending on the flags you use to launch the VM <clarkf>virtio drives are /dev/vd* rather than sd* <dariqq>i just found out that one can nest with-imported-modules/with-inputs , i just added a shepherd service that is provided by my own package. That is pretty cool <PotentialUser-42>I can just boot on the first config, the initial one, but I don't know on which disk it boot <clarkf>are you modifying the config at all? or just pulling and reconfiguring? <nckx>Rutherther: Does ‘source foo.sh’ even work under fish? (Maybe for a simple KEY="value" shubset?) <resu>ok so im trying to import my stow config to guix home <resu>and it says home-dotfiles-service-type is unbound.... what module am I missing I thought it was part of (gnu home services) <clarkf>(gnu home services dotfiles) i believe <PotentialUser-18>gah, i disconnected. dunno if anybody gave me advice (im person with `"gkr-pam: unable to locate daemon control file` issue) <resu>I just did a "guix home search" and figured that out <resu>still learning to navigate guix atm <vagrantc>no need to be so harsh, you're learning :) <Rutherther>PotentialUser-18 I dont know if anyone replied, but just so you know for next time, this chat is logged, URL is in motd. So you can find your question and potentially answers there <PotentialUser-18>i added debug logging to gdm3, switched to x11 (it fails more gracefully than wayland). I'm not convinced that `gkr-pam: unable to locate daemon control file` is the thing thats broken - it seems to respond after and still go thru login stuff <PotentialUser-18>i also have a gnome `JS ERROR: extension [...] failed to look up menu_file for "applications.menu"` error <PotentialUser-18>i installed glibc-locales with guix, (not guix home) and that error went away. gnome is still broken tho. <PotentialUser-18>*sigh*, guess ill reinstall... again, and see if i can narrow down what breaks it. <PotentialUser-18>journaling this because this is the 3rd time ive had to reinstall because guix (i assume) broke something in the base system and im frustrated :/ <PotentialUser-18>installed debian bookworm, pulled latest updates, installed gnome-core, rebooted. <PotentialUser-18>ran 'guix pull', sourced the profile (as suggested by guix pull), ran `hash guix` <PotentialUser-18>ran `guix pull` again (because now its pointing to the guix binary installed by the system guix binary). <Rutherther>PotentialUser-18: when using guix on foreign distros you have to make sure to source the profile yourself, until you do that, hash guix won't have any effect since you don't have the location guix pull puts the profile to in your PATH <PotentialUser-18>i rebooted and it defaults to the system guix, which breaks guix home since "module not found", i am going to uninstall the system guix binary <Rutherther>PotentialUser-18: so you have the source now in some kind of .profile file? <PotentialUser-18>ugh, the root guix profile is now pointing to the $USER guix profile, because I su'ed to root before installing <Rutherther>PotentialUser-18: so then it makes sense you don't have your user's one after reboot, you need to always source the profile first <PotentialUser-18>Rutherther - im not sure i _should_ have to do that, /etc/profile.d/guix.sh should source the profile in the user's home dir <PotentialUser-18>ugh, now im just annoyed that the root profile is pointing to the user's profile <PotentialUser-18>do i just whack the entire guix install and install from the guix foreign script? <ieure>PotentialUser-18, It defaults to the system guix when you reboot because you didn't add the profile sourcing bit to your shell dotfiles. <ieure>Maybe I'm wrong? But that seems like the problem. /etc/profile.d/guix.sh be damned.