IRC channel logs
2025-01-29.log
back to list of logs
<coyotes4ys>gnu/services/base.scm:769:37: error: host name 'propeller 2' contains invalid characters <freakingpenguin>coyotes4ys: See assert-valid-name in gnu/services/base.scm. No whitespace allowed in host names. <coyotes4ys>freakingpenguin, i don't have gnu/services according to my terminal <ieure>coyotes4ys, That's in the Guix source repo, which will be in ~/.cache/guix/checkouts <ieure>Rutherther, Huh, does Guix not let you boot into admin mode to fsck like Debian? <ieure>Have had plenty of times when some system or other wouldn't boot and I was able to fsck without needing any other media. <PotentialUser-47>Has anyone packaged or tried to package Niri before? I've checked the Toys/Webring thing (sorry I don' <PotentialUser-47>Has anyone packaged or tried to package Niri before? I've checked the Toys/Webring thing (sorry I don't know the proper name) and I tried to use cargotoguix but it generated a 7.2k long file and I'm a bit intimidated. <coyotes4ys>so if i want to configure openbox by editing the rc file, i need to copy it from gnu/store/... to home? <coyotes4ys>i figured that out/remembered, but now i have a diff question <coyotes4ys>if i change the resolution in xrandr, of course my words will appear bigger, but will it affect videos and games too? <coyotes4ys>does anyone know the package that has the alt+f2 run command thingy? <vhns>I just realized something, gnu/system/ldap.scm defines backend-userroot-configuration but never exports it. The docs refere to it as well. Someone must've misstype as what gets exported is backend-configuration instead <vhns>(which is never defined within the file) <PotentialUser-47>I'm having trouble packaging a hyprland plugin called hyprscroller. When I run `guix build` I get an error when `make install` is run. I've tried to fix this by just removing the installation procedure, but that didn't work. <vhns>PotentialUser-47: I assume the issue to be that you're using cmake-build-system despite it being a plain make project. <vhns>PotentialUser-47: could you try using gnu-build-system instead? <PuercoPop>PotentialUser-47: the error is that there is not install 'rule' in the generated Makefile. You'll probably need to modify the build phase that calls make install. <PotentialUser-10>vhns I see, the project does have a CMake file, so I guess I got confused. How would I go about using just make? The gnu-build-system seems to want a ./configure `./configure` <vhns>You can just delete that step. <PuercoPop>But the Make file takes care of 'operating cmake' <PuercoPop>With gnu-build-system you'll have to delete the configure step and add cmake as a (build time?) dependency <PotentialUser-10>Alright, so I switched it to `gnu-build-system`, removed the configure step, and added cmake as a dependency. It's now complaining about not having permission to create a '/.config’ file. I assume I have to patch the Makefile? <PuercoPop>PotentialUser-10: can you paste the contents of this file: View build log at '/var/log/guix/drvs/f8/psc32gic6xvvps7z4aqpx2xv8qqffv-hyprland-plugin-hyprscroller-v0.47.0.drv.gz'. <PuercoPop>Btw I just checked nix and they do use the stdenv (akin to gnu-build-system) instead of the cmake build system to build the plugin and add to cmake as a nativeBuildInput (build-time dependency) along with pkg-config <meaty>Is there anything special that needs to be done to run appimages in guix? <meaty>Whenever I try to run one I get bash: ./[...].AppImage: No such file or directory <coyotes4ys>why is my installation taking nearly 100% of my processing power? <meaty>Hi, I can't get an appImage to run in guix 'cos it expects ld-linux-... to be at /lib64/, and not in the store... how do I get the system to run it? <coyotes4ys>i mean my guix install. and it's been hung on installing gcc for quite awhile like maybe 30 mins <coyotes4ys>closing everything except tint2 didn't help it seems <meaty>coyotes4ys: did you just pull? a substitute may not have been built on the upstream servers yet, so your computer is compiling and testing gcc <coyotes4ys>man this sucks. it's been installing qbittorent and dependencies for 1.5 hours now and i'm going to lose wifi soon <coyotes4ys>i had problems installing several things in a row regarding substitutes so for this one i did -nosubstitutes or whatever the option is. i don't remember exact syntax but i did it right <meaty>it's a major shortcoming of guix <PuercoPop>PotentialUser-10: the issue is that the make rule is tyring to call `xdg-user-dir` which available. You need to add it as dependency <meaty>coyotes4ys: "substitutes" are precompiled binary packages <meaty>turning them off means your computer compiles EVERYTHING <meaty>everything that was built stays in the store, if it was in the middle of compiling it disappears <meaty>the good part is that whatever was compiled should stick around until your next 'guix gc', so it's not a total loss coyotes4ys <meaty>please somebody tell me how to run appimages in guix <squid64>After a day of using EXWM I can say I like it. Was one of the choice in the guix installer. I wasn't ever a fan of tiling wm but now that there is one that uses Emacs, it's actually not too bad. <squid64>Also yeah I haven't been able to run appimages myself either <meaty>'file' says that /lib64/ld-linux-x86-64.so.2 is the interpreter, but guix ovbvi doesn't have a /lib64/ dir, that file is in the store <meaty>so how can i patch the appimage to use the right "interpreter" <meaty>the nuclear option is to go into sudo and create a symlink, but there must be a better way <wakyct>can you export the ld library path and then run it? <PotentialUser-10>PuercoPop, so I added `xdg-user-dirs` as a dependency, but now I'm getting this `mkdir: cannot create directory ‘/homeless-shelter’: Permission denied` <PotentialUser-10>Google says that it's related to `$HOME`not being set, not sure what I would even need to set that to. <PuercoPop>Yeah, that is part of 'hardnining' guix and nix do the avoid building a packages modifies anything other than the store <PuercoPop>Ok, so what the script is trying to do is place the plugin in the ~/.config/hypr/plugins. <PuercoPop>How do hyprland plugins work? You could remove that step and add a home service that symlinks a directory there to the output in the store? <PotentialUser-10>All you need is the path to the .so file for them to work. So something like this would work `hyprctl plugin load ~/.guix-home/profile/lib/hyprscroller.so` <PuercoPop>Unsure what the best way to handle this is. Maybe to a home service to symlink? On think it seems that each plugin takes the hyprland derivation/package as argument. <guestuser1>Question: what image for Advanced / Expert Install ? does it include non-free? possible to select DE / WM ? can you explain about Scheme/Guile ? Thank you! <tae>Is there a recommended way to read docs from Guix, something like "guix info" <PuercoPop>tae: texinfo docs are better read either in the browser or using Emacs (C-h i) <eikcaz>I'm working on a guix service definition to configure Syncthing. Since I think people should be able to easily reference Syncthing documentation to configure via Guix, I want record names to match the corresponding entries in Syncthing documentation exactly, but Syncthing uses camelcase <eikcaz>Is this a valid exception where camelcase is OK in the Guix repo? <eikcaz>e.g., to configure the insecureSkipVerify section of ldap, I named the corresponding record field ldap-insecureSkipVerify. "insecureSkipVerify" is also how it must appear in the generated config file <futurile>seems sensible, removing as much dissonance between the underlying applications docs and using it in Guix - no idea if it's 'allowed' though heh <futurile>eikcaz: you should add your voice on the Codeforge migration proposal - Thomas has some good questions in the guix-devel thread <sleepylkw>is it possible to run guix system init from a non-guixSD system? trying to do so but starting cow-store with herd gives me a missing shepherd socket error. <Rutherther>sleepylkw: of course it is possible. As for the cow store, are you on a live iso or not? the cow store is only making it so that stuff from the store is backed by disk and not ram. Your ram might not be large enough to hold the whole system. If you are using live iso, you can just call overlayfs manually instead of making cow-service (that's part of the guix system iso) call it for you <Rutherther>this is to manually execute what cow store service would do: "guix repl --expression="((@@ (gnu build install) mount-cow-store) /mnt\" \"/tmp/guix-inst\")"" <futurile>sleepylkw: maybe do it in a VM if it's giving you trouble? (that's my easy solution to everything heh) <carloratm>Hi all, I have a very n00b question from someone who just installed guix. The manual suggests to do a guix pull after installing guix on a foreign distro, I did that and it took 1.5 hours on my laptop from 2014. Is that the norm when running guix pull? <willcl-ark>I am trying to run guix on NixOS, and it mostly seems to run OK as root user, however the moment I try to run as another user, I see "spurious SIGPOLL" and "unexpected build daemon error: interrupted by the user" in the journalctl logs for the service, and the command (guix pull) does not complete successfully. Does anyone know what this might indicate is going wrong? guix also does not seem to want <willcl-ark>to create a profile for that user, which may be a cause or an effect... <PotentialUser-19>Hello, I'm trying to install guix via a virtual machine as explained in the guix-handbook at point 12.17 (german). First I installed qemu via MSYS2. Than I downloaded the qcow2-Image and followed the handbook. The first issue I got ist that the hardware acceleration doesn't work: -enable-kvm -m 1024, -enable-hax or -accel kvm, -accel hax generate a <PotentialUser-19>message like "invalid kvm accelerator" or "invalid option". As hardware acceleration is not necessary, I tried without it, but it ends with a kernel panic message:"not syncing: system is deadlocked on memory". My processor: Intel(R) Core(TM) i5-6300U CPU @ 2.40GHz 2.50 GHz, ram: 8,00 GB, System: 64bit. I can't find what is wrong. Any idea? <Guest83>Hello all. First-time guix user. I just installed the guix system on my laptop. Reading the info pages it states: "Likewise, to _upgrade_ system <Guest83>software, you first fetch an up-to-date Guix and then reconfigure your <Guest83> sudo guix system reconfigure /etc/config.scm" <Guest83>Could somebody please clarify if it's possible to "guix upgrade" without first "guix system reconfigure", and if I did so, would I then have to run "guix system reconfigure" later to persist the change? Many thanks <Rutherther>Guest83: guix upgrade and guix system reconfigure are unrelated things. With guix upgrade you upgrade your user's profile where you install packages with guix install, whereas guix system reconfigure applies your config, which contains the system profile that will be upgraded as part of it if guix updated. No, you do not need to run guix system reconfigure to upgrade your user profile <Guest83>Rutherther, thank you. So for package upgrades it's "guix pull" -> "guix upgrade". Apart from when changing my config, when would I want to run system configure? <Rutherther>when you want to upgrade the packages installed from your system configuration <Rutherther>which might be the basic system stuff, and also all the system services, which you probably don't want to get old, since there could be vulnerabilities <Guest83>Ah ok. So it sounds like I need to read more on the separation between the user profile and the system configuration and services <Guest83>I guess the key phrase in the documentation I quoted is *system* software <futurile>carloratm: no that's not the norm. It's not *fast* for sure, maybe 10 mins in total to update everything. It has to download the archive (which is a git repo) and there have been problems with the GNU infrastructure as they're being attacked. After that it downloads the binaries (called substitututes), which it gets from other servers <Altadil>carloratm: the very first guix pull is significantly longer than normal, because there has been a lot of updates since the installer was last updated. New ones should be faster. <carloratm>it's a bit annoying for newcomers I suppose, maybe a sentence in the manual would help <Guest83>I was also in the same boat as carloratm. The info pages suggest running a guix pull and then running a system reconfigure. A comment that the first run may take longer than subsequent runs would certainly be helpful for a new user <PotentialUser-19>ok, I did it and I guessed right: Without memory no emulation. kvm was the wrong accelerator. I found the right one in the qemu documentation under point virtualisation accelerators. The right one for my system ist not kvm but tcg. In MINGW64: -accel tcg THX =D <jonsger>apteryx: AFAIK, the bank account was closed and opening a free one took a lot of time. In the meantime the money of the Foundation was on Andreas' private bank account. He did also payed for expendings during that time... <apteryx>I wonder what happened, if anything, since. <csantosb>carloratm: out of curiosity, how do you install guix ? using your distribution package manager, or the binary installer ? <csantosb>New users install guix as any other software: using their local package manager. After a first guix pull, they just give up. <csantosb>carloratm: keep on going, it's clearly worth it. Today, I'm a happy guix user in a similar situation. <carloratm>I've been using nix to manage development environments at work for the past year. Being a gnu-head I would love to replace it with guix shell <jonsger>apteryx: I think we have a working bank account of the Guix Foundation again, for a few weeks now <csantosb>carloratm: Welcome abroad ! Don't hesitate to share any difficulties here, willing to help <ieure>Sigh, Savannah's hosemode yet again: `guix pull: error: Git error: unexpected http status code: 502' <attila_lendvai>so, i'm reading the survey blogs, and wanted to record some typo fixes... but apparently i cannot pull the guix repos, i get error 502 from git pull. is that only me? (i'm on a strange connection...) <ieure>Yeah, I just mentioned, Savannah is horked. It had the same problem yesterday, not sure if this is the same one or what. <ieure>attila_lendvai, The web page is a different thing than the Git service, I think. <attila_lendvai>ieure, well, i'm pulling through a https link... but probably there's some magic under the hood <ieure>attila_lendvai, No magic, likely a reverse proxy that delegates to different backends based on the path of the request. <ieure>Just typo'd a command and now I kind of want a real `guix pulp'. <attila_lendvai>so, i'm recording this commit with typos in the blogpost, and i can't help having this lingering feeling that it'll be wasted effort and it'll bitrot in the patch queue indefinitely... <ieure>csantosb, `sudo dd if=/dev/urandom of=$(blkid | cut -d: -f1 | shuf | head -1)' <csantosb>attila_lendvai: new codeberg infra is expected to help with this respect <attila_lendvai>and the worst part is that it's demotivating me to even read the survey blog posts, because then i'll skip over the typos, and things will remain in a state of not-beautiful... <csantosb>I just wonder how really huge projects (linux kernel?) handle this kind of situations <jlicht>it helps that generally folks are paid to work on the Linux kernel I guess <attila_lendvai>i think the lack of (officially admitted) central authority for guix is a source of more problems than is generally considered <attila_lendvai>and i found it confusing as a newcomer that i had no idea whose voice mattered how much when i started to engage the community. i.e. for too long a period i had no idea whether i'm talking to one of the original authors here, or just a random ircnick like myself... <ieure>Savannah seems to be unhorked now, fwiw. <wakyct>attila_lendvai I understand where you're coming from but to me the communal feeling of Guix was one of the main draws, after running Manjaro <attila_lendvai>wakyct, but after being around for a while, do you still believe that that feeling is in harmony with the observable reality? because i seem to have mapped the authorities, at least to some extent, it's just not explicit. and there's nothing wrong with that, because no project like guix couldn't function without black and white commit bits and (competence based) authorities making some of the decisions... <elevenkb>What is considered a bug in guix? Rust-analyzer wasn't completing stuff from std, even though it could complete stuff from the workspace. I think it's because we don't build the proc macro server "rust-analyzer-proc-macro-srv" but I have no idea. For now since I'm a CS beginner I think I'll just stick to nix for those projects that seem to need it until I have a more comprehensive. <elevenkb>sorry incomplete sentece... comprehensive understanding of CS. <elevenkb>Should I report my issue on the bugs-guix mailing list? <ieure>Hmm, looks like golang is broken on aarch64-linux. <ieure>Because it's trying to bootstrap with x86_64-linux binaries, whoops. <ieure>When I `guix describe', I get shown the current generation, corresponding to my last `guix pull'. How do I see all generations? `guix describe --list-generations' gives an error, `guix package --list-generations' tells my my per-user profile doesn't exist (which is the case because I'm using Guix Home). <ieure>`guix home list-generations' doesn't show what I want, because I haven't `guix home reconfigure'd yet. <look>ieure: how about 'guix pull --list-generations'? <theesm1>just a heads up for those being in Brussels for guix days already: au bon vieux temps is closed today (and probably the next days as well as it seems) <jlicht>good trial by fire for the community consensus procedure if we want to congregate elsewhere ;) <tschilp>It's guix 25651bab10800cf0, my system reconfigured successfully already, it's just home stuck. Does rebooting make any sense here? <Rutherther>tschilp: stracing guix commands will not going to give you anything if it's not the guix command itself failing, but a derivation build failing <tschilp>Rutherther: thanks for the hint -- do you have a suggestion how to dig in there? <Rutherther>I don't know where lfam gave you that advice and why. But if the check is failing because of a timeout, you could enter the build environment and run the tests themselves with strace. That could give you an idea if something is happening or not (meaning it's likely stuck) <tschilp>Ohh, quite likely I misunderstood lfam's advice :) I just ran it against the ~guix home reconfigure...~ <Rutherther>running strace on guix home reconfigure cannot help you, because the build is being ran through the guix daemon, you are just seeing the log from it <tschilp>Could you point me in a direction how to enter the build environment? I do not know much more than ~guix [home|system] reconfigure $MY_FILE~ and ~guix build $SOMETHING~ <Rutherther>tschilp: 1. you can run guix build pantalaimon --keep-failed. That will do the build normally, and upon fail, there will be a directory in /tmp with the build. Then 2. you can do guix shell -D pantalaimon, that will give you a shell with needed dependencies, 3. you source the environment-variables file in the build directory, 4. you go to the source directory and run what the build was running, probably something like "python3 ./setup.py test" <ieure>tschilp, Tried building pantalaimon on my machine, I get some test output, but it also hangs. <ieure>"tests/pan_client_test.py::TestClass::test_login[pyloop] PASSED [ 5%]" <ieure>Doesn't seem to proceed further. <ieure>It sounded like you were getting no output at all, though? <tschilp>Rutherther: great, thank you a lot for this! I will give it a try. <tschilp>ieure: yes, I get a path to the build log, but it's empty. When I tried yesterday, it turned out to be some faulty file with ~missing EOF at the end of file~ or something quite similar. <tschilp>I'm just running ~guix build pantalaimon --keep-failed~ and I'm at the ~tests/pan_client_test.py::TestClass::test_login[pyloop] PASSED [ 5%]~ point as well, my cpu does goes rather quietly though. <Rutherther>tschilp: you can probably just C-c it then, because otherwise you will be waiting for the timeout <Rutherther>even if you do that, the folder will stay in /tmp <ieure>tschilp, Yeah, this is an annoying one. Probably the right move is to run `guix build -S pantalaimon' to get the source, then `guix shell -D pantalaimon', unpack the source, and manually try the build steps in the package definition. <ieure>The build log isn't going to have anything you don't get with `guix build pantalaimon'. <ieure>And since the test is timing out, `guix build -K' isn't much help. <Rutherther>ieure: it is, because it does the steps before, so you can just execute tests directly, without all the previous steps <tschilp>I'm just waiting for the build dependencies to arrive, but need to get some food now. will report a little later! <orahcio>Hello, if I create a channel with some updated python package, how to choose that version of my channel instead of the guix one? The name is the same, it just changes that version <ieure>orahcio, It depends. How are you currently getting python, just `guix install python'? <orahcio>ieure: I mean just a pypi package python-pyglm for example <ieure>orahcio, Sure, same question applies though. <orahcio>I use python-pyglm@newversion on inputs and I guix ask for the module <Rutherther>orahcio: if you are writing guile code that uses the package, you should use-module the module it's in, and reference it with the symbol <orahcio>Rutherther: I need to rename (define-public python-pyglm@newversion intead just python-pyglm? <Rutherther>you don't need to do that, but you can put the version to the name if you want, it will make things clearer. Though @ is probably not allowed in symbols, or at least - is used usually <orahcio>Rutherther: when I choose a version with `guix install` I choose with at @, but if I want to choose the version on inputs in my .scm file? How to I proceed? I can not use the same name? <Rutherther>orahcio: what you do with guix install XYZ@a.b.c is different. That is a string specification of a package - a filter saying give me package with name XYZ, a.b.c. In guile code you reference specific guile symbols, you aren't using specifications <orahcio>Rutherther, inside a input field I just use the name on define-public, I can not use @a.b.c after the name, so I need to rename that updated package <Rutherther>orahcio: it's not true that you need to rename it as I've stated previously. I don't know what you are trying to say with this message though <orahcio>Rutherther: I try to use (propagated-inputs (list python-pyglm@a.b.c... and guix ask me for the module with python-pyglm@a.b.c, but I am using -L flag to set my channel after `guix build` <Rutherther>as I was saying, you're referencing symbols, not using specifications... <Rutherther>I really don't understand what is not clear about that <Rutherther>the "(define-public my-symbol" defines a symbol called my-symbol, so you need to specify my-symbol, after getting it in scope - ie. with use-modules, or just referencing directly with (@ (module) symbol), but you don't typically do that <orahcio>Rutherther: Yes, you was very clear, I think my way to think named it as "change the name", anyway I could see if I use the same symbol, guix chooses the new version <bostx2>Anyone in Brussels today? The pub "Au Bon Vieux Temps" is closed <bostx2>I meant the pub Ludo wrote about on the mailing list (Guix Fringe Days - FOSDEM) <ieure>If anyone is interested. Not my domain and don't have time to investigate further at the moment. <ieure>Oh... looks like issues.guix.gnu.org hasn't updated in four days. sigh <lilyp>ieure: you can get the bugs by their number, it should be just the front page that's stuck <divya>I wish I could be at FOSDEM this year. I hope next ywar <divya>Those attending, can someone give a rough estimate of how much it costs, in terms of staying and all? <divya>I think flight costs alone to Brussels would've been 300€ or so for me. <ekaitz>divya: it's not superexpensive comparing to your flights <divya>ekaitz: Cool. Either I'll save that extra money for it next year or pester my employer for it :) <divya>Must be supercool to be among the hackers! <ekaitz>divya: just presence is not enough for having a good time <ekaitz>fosdem has a lot of that, but not a lot of human connection <divya>No? Oh, I thought since they separate smaller rooms, so that might lead to closer communication <ekaitz>it's still a experience to have and a good place to give a talk <ekaitz>and you can meet people in a smaller context out of the event itself <divya>ekaitz: In your experience, which conferences/meetups have been the most engaging and communicative? <ekaitz>divya: always local and few people :) (Guix Days is cool btw) <divya>ekaitz: Right, I forgot it's a FOSDEM _fringe_ event. <tschilp>mhm, when I enter the ~/tmp/~ directory I receive from running ~guix build pantalaimon --keep-failed~, get the dependencies by running ~guix shell -D pantalaimon~, source the environment-variables file, and run either ~python3 setup.py test~ from the source directory, or ~pytest~ from the source/pantalaimon directory, no tests will be run (but also no errors issued). But if I create an ~installation~ directory in the source directory, <tschilp>there's quite some errors about the pythonpath, namely that ./installation/[..] does not support .pth files -- bad install directory or PYTHONPATH. I ran a rather desparate 'export PYTHONPATH=$GUIX_PYTHONPATH:[...]/installation' before running the command again, but I guess that's the wrong way, as it results in the same 'error'. Good night guix!