IRC channel logs
2025-03-27.log
back to list of logs
<meati>would anyone be willing to run `guix shell nicotine+ -- nicotine` for me and LMK what it says <meati>it's segfaulting on my machine, which is a regression <meati>or, would be, if it's happening for others, i guess <old>meaty: no segfault. I get a prompt <old>tested on: guix d0510dc <meaty>old: word. I'm on ce086e3, so if this is anything it's newer <meaty>hmm... strange. Nicotine works fine in a shell but segfaults outside of one <meaty>`guix gc -verify` comes out fine, `guix package -u nicotine+` says "nothing to do"... <meaty>using a shell partially fixes Nyxt too... could by store be broken in some hidden way? <meaty>guix gc --verify=contents comes out fine... what could be happening <RavenJoad>meaty: More likely what I would suspect is that a shell has additional wrappers present in it (because it is a full profile) compared to an installed package. Why? I don't know. But that would be my first question. How does a package made available with guix shell vs guix package -i differ? <RavenJoad>Unless you have manually messed with the store, I doubt you could have screwed it up in a way that managed to make it so inconsistent that guix shell and guix package behave differently. <RavenJoad>sneek: tell civodul I will soon get around to putting together a patch series together for the Xen stuff. I would appreciate him looking at it when it is done. <sneek>civodul, RavenJoad says: I will soon get around to putting together a patch series together for the Xen stuff. I would appreciate him looking at it when it is done. <meaty>sneek: tell RavenJoad I was able to get the packages working (the same way they ran in guix shell, anyways) by removing them from my home config and reconfiguring, then installing them via guix package -i. I tried switching back to confirm, and found that they didn't work again.... I don't know what to make of this. <sneek>RavenJoad, meaty says: I was able to get the packages working (the same way they ran in guix shell, anyways) by removing them from my home config and reconfiguring, then installing them via guix package -i. I tried switching back to confirm, and found that they didn't work again.... I don't know what to make of this. <Rutherther>meaty: my guess would be you didnt relog to get the new search paths <meaty>Rutherther: am back, logging out then in again didn't make the home-profile-installed packages work <meaty>unless you meant something else? <Rutherther>I meant after every change you did before. You seem to be having different behavior for home and user profile, that is usually not a thing unless search paths are involved <ss2>I have a question already: How does one reactivate dormant tty sessions on a server? I have one and would rather do some work on it before having to reboot it. <ss2>Restarting term-ttyX or term-console doesn't help. The screen stays blank. <civodul>ss2: what do you mean by dormant sessions? <ss2>civodul: Sorry, messed my wording again. If no screen is attached, the video output goes off eventually. I'm looking for ways to get it back on again. <ss2>There's just no signal. I can log in though. <Venoflux>Hello folks! I am still trying to get the hold of guix as an emacs package manager, I am using a manifest file but I wonder if it is the way or there is a recommended way of using it (guix on top of debian). Also how can I use the --with-input=emacs-minimal=emacs in the manifest.scm or any other way to keep it as a default? <ss2>Take note of the last section. <Rutherther>Venoflux: look up options->transformations. You can definitely use them like that in guile <futurile>PotentialUser-63: Guix packages are fully buildable from source. So you can build every package yourself. Many users will use 'substitutes' which is the binary package built by the CI system - so they don't have to build everything themselves. <futurile>PotentialUser-63: because it's a functional package management system (the same inputs get you the same build, and signature on the build) a user can also check that a package they build is the same as the one CI built - so it's a way of challenging that the binary package you received hasn't been backdoored <PotentialUser-63>I'd rather not do gentoo again, as it was once fun but now I just rather download and go kind of thing <futurile>PotentialUser-63: yeah, exactly - that's why most users will actually use the binary packages <futurile>PotentialUser-63: but it's nice that you can "check" the binary package, and if you do want to compile things, or change a package, you can <civodul>ss2: oh, weird; that may be related to the video driver, no? (i don’t remember experiencing this) <ss2>civodul: true. I will investigate further. <jA_cOp>A couple of my custom shepherd services seem to get restarted on shutdown, and then shepherd waits forever for them to stop. I guess because they respawn by default and maybe have a non-zero exit code when being killed by shepherd? I do want them to respawn, but not if shepherd is the one that killed them in the first place <debbugs>New or unarchived bug in guix-patches: "[PATCH] gnu: ruby-3.1.7: Upgrade to 3.1.7 [fixes CVE-2024-{27280, 27281, 27282}, CVE-2025-{27219, CVE-2025-27220, CVE-2025-27221}]" https://issues.guix.gnu.org/77304 <abralek>I have some patches for hostapd. dynamic vlans configuration in particular and I would like to write some tests using mac80211_hwsim. It seems it is not enabled now. There are also other debug options that could help to test things. Not sure, would some linux-libre-kernel-test-suite be ok? <abralek>I now I can do it in my channel, but that would mean no tests in guix repo) <divya>Would a commiter please look at this reviewed patch series and merge it? <divya>ekaitz: Would you be interested in looking at this? It's a nice package to have for low-level devs <ekaitz>divya: holy shit, that's a lot more than I can handle right now <divya>Oh wait my browser didn't load the last few messages <jA_cOp>to answer my own question: I have to make my shepherd services depend on 'user-processes <Rutherther>Can I make a derivation that will be 'ungraftable' so that when grafts are applied to something referencing it, this one and its deps are skipped? <df>what determines whether a service can be auto-restarted after a reconfigure? <df>for example, it seems like it would be pretty safe to restart shepherd timers, but none of mine are <df>and in fact it seems like the majority of non-one-shot services are not (including the timer service itself) <df>some of them are obviously understandable, you don't want your desktop restarting under you, and I wouldn't want my getty terminals restarting if I was using them <df>there are also quite that I don't know the purpose of yet, like the file-system- services (which aren't one-shot?) <Rutherther>File system service stop is used to unmount the filesystem <df>Rutherther: oh, I figured that would be a one-shot <Rutherther>As far as I know one shot services are stopped after they start, meaning that nothing would stop them on shutdown <df>hmm ok I do still need to read the shepherd manual, but you're saying basically if something needs to do stuff at both at init and shutdown it won't be a one-shot? <Rutherther>As far as I know one shot services are stopped after they start, meaning that nothing would stop them on shutdown <df>LOL, I thought that urandom-seed would be safe enough to restart, but when I did so my desktop took a nosedive and I had to reboot <df>on the bright side I don't need to restart any more services <df>and if there was a kernel update it'll have been applied <ampinga>hello, does anyone know how to install node modules/packages globally on Guix system? I tried "npm install <node_module> -g" but it doesn't work. <ieure>ampinga, Guix doesn't really have a notion of "globally installed" in the way that stuff like NPM (or Python tooling) expects. The closest thing to that would be to put the Guix packages for those node modules into your operating-system configuration. But that's a bad idea. <ieure>ampinga, Because in 99% of cases it's the same effect as installing into your user profile, but is more work. <ieure>(Assuming you're asking "why is that a bad idea?") <Rutherther>ampinga: you need to set your npm prefix to an existing folder, ie. "npm config set prefix '~/.local/'". But if any npm program needs an so library, you will have a problem, the only stuff that will work is npm packages just using the javascript <Rutherther>guix system isn't fhs, libraries aren't exposed to downloaded programs / binaries <ampinga>Rutherther: oh I see so there's a counterpart for every choice. <ampinga>Rutherther: Sorry, maybe I misunderstood. I just meant that there were drawbacks even in the solution. The thing with the SO library. <Rutherther>'so' refers to any dynamically linked library on linux <Rutherther>so I don't understand what you mean by _the_ so library <ampinga>Rutherther: Oh I see, sorry. I really misunderstood. <ampinga>the globally installation works fine now thanks. <nick_1>Hola, just few questions, what type of installer Guix use? Ncurses, GUI? what options for beginners? use for Developer or Server Administration? what Tools you use often ? Coreutils? what about minimal iso, does it contain network-manager? Non-free? <Rutherther>nick_1: it's tui, not sure if ncurses, could be. Guix has coreutils. Minimal iso doesn't contain network manager. Official guix channel doesn't contain anything unfree <bjoli>I am getting an error about guix system reconfigure is run with an old version of guix and that I need to run a guix pull. This does nothing. <Rutherther>and what guix instance are you using? type -p guix <bjoli>It started after a system reconfigure that produced a non opting kernel <bjoli>Givve me a couple minutes, my daughter just woke up <bjoli>Rutherther: ~/.config/guix/current/bin/guix <nick_1>Rutherther: thank you for the answer , know about tui and its options? what iso include nmtui? what iso for minimal window-manager example: LXDE? what about pos install? btw do you use something like Gnuzilla / Hubzilla? <Rutherther>nick_1: there is no other iso supplied other than the minimal one. I don't know options of the installer, I've never used it. You can of course configure a display manager post install. No, I don't use gnuzilla nor hubzilla <Rutherther>bjoli: okay, and what was the output of guix pull? <bjoli>Ruth: the usual followed by nothing to do <Rutherther>bjoli: so that explains it then, if it says nothing to do, it did nothing, meaning guix system reconfigure will still tell you it's old guix <Rutherther>bjoli: and that guix pull did nothing is most probably caused by pinned channel <mirai>what do I need to do to solve "error: luks-device-mapping-with-options: unbound variable" <nick_1>Rutherther: minimal iso contain drivers? its offline installer? <bjoli>Rutherther oh. The first pull did something. It didnt help. Subsequent pulls do nothing <Rutherther>nick_1: it is not really an offline installer, though you could probably manage to install offline by it if you used the default config, because it should have everything it needs for a basic config. Not sure if the tui installer can respect that, so you might have to do manual install <Rutherther>nick_1: I don't know what you mean by 'contain drivers', it is running linux libre so it has all drivers linux libre has... <Rutherther>bjoli: and how are you running guix system reconfigure specifically? <nick_1>Rutherther: thank you for you reply <bjoli>Rutherther guix system reconfigure path-to-file-that-works-fine-except-for-me-borking-it-with-a-failed-mount <nick_1>have you tried something like Trisquel, Parabola? <Rutherther>bjoli: hm, that's strange behavior then. And guix describe will show you generation with today date? <bjoli>A guix pull I did just now did download something. The system reconfigure is still not working. Guix describe shows march 27. The failed system reconfigure was two days ago. <nick_1>Rutherther: do you know Guix compatibility compared to others? <Rutherther>bjoli: guix system reconfigure not working? That's new info from you, that message you sent earlier is just a warning, not an error <bjoli>Oh. It says the parameters in var something something is outdated. I need to run guix pull, which does not change the error <Rutherther>bjoli: please send the log from guix system reconfigure <bjoli>(sorry, left my computer. Daughter waking up again). Give me some minutes <bjoli>Can I run it verbosely? Or do you just want the output from the command? <nick_1>Rutherther: with network , btw what Provider do you use? <Rutherther>nick_1: sorry, I have no idea what compatibility with network means, and also no idea what type of provider you're talking about <bjoli>Ruthether: guix system: error: unrecognised boot parameters at '/var/guix/profiles/system-5-link/parameters'. <bjoli>Then a hint about guix probably being outdated and that I need to run guix pull. <Rutherther>bjoli: that hint is just automatic hint, it doesn't have anything to do with the error.. please next time send the error first, not the hint <Rutherther>bjoli: as for the error, is the file empty? or how does it look like? <nick_1>Rutherther: internet provider, by network compatibility was asking about drives / modules, compared to Debian / Gentoo / Guix, its best option? <Rutherther>bjoli: your files are corrupted, that is the problem. You will need to get rid of the corrupted files. You can try "guix gc --verify=contents,repair", but it will likely fail and you will have to delete the generation and gc it. <Rutherther>bjoli: please before you do start doing anything, could you send the guix revision from /var/guix/profiles/system-4-link/channels.scm? <Rutherther>(and /var/guix/profiles/system-5-link/channels.scm as well) <Rutherther>there is/was a bug that causes file corruption, I need to check if your revision was old enought to be expected, or that you're affected after commits that fixed the behavior in my reproducer <Rutherther>nick_1: I am using a local internet provider that covers like 50 km radius, so I doubt you knowing it would help you in any way... I don't know what to tell you, because I have no idea what options Debian nor Gentoo offer. The official Guix channel uses linux-libre with no non-free parts, meaning no non-free firmware. So look up the chips you're interested in and see if they are supported with linux-libre and free firmware. <Rutherther>nick_1: of course you don't have to limit yourself to the official channel and can use full linux with non-free firmware. That is just not the official way, and please discuss it in #nonguix, not here <nick_1>Rutherther: what you mean by full linux with non-free firmware? its Guix fork? non-official? <Rutherther>no, not a fork, a channel. Guix has channels that give you more packages/services, you might use multiple <pegaso>like repositories in another distros <nick_1>can you please tell about Guix Userspace / Kernel? your choise GUI or TUI? what packages you use often? <bjoli>Rutherther: nope. I run on a ups. <Rutherther>this isn't making me happy :( could you also send your config? <nick_1>can you explain tell me about Guix configuration / hardening? know about its Privacy / Anonimity options? <pegaso>nick_1: I use standard gnome with wayland and things like rgb, dino for xmpp, icecat for browsing (actually a big reason for me to use Guix) <nick_1>pegaso: got it , what type of configuration? custom? <Rutherther>nick_1: if by hardening you mean something akin of systemd hardening, then guix doesn't do that much, at least not yet. There is a wrapper to run services in separate namespaces, but it is not used much <pegaso>nick_1: for hardening you mean? I just use the standard configurations from the devs which probably know more than I do. I'm just a normal joe <emacsomancer>is there a defacto or prescribed file-name and location/path for system `config.scm` and guix home `home-config.scm`? <Rutherther>emacsomancer: no, you always give the name in the flag, so it's completely up to you, no location is 'preferred' <ieure>emacsomancer, I keep mine in Git repos in $HOME. <identity>emacsomancer: i just keep them in ~/src/{home,system}.scm (inside a git repo) <emacsomancer>so no standard suggestions in any the docs? (I'm trying to run an update helper, and was just wondering if it made sense to have defaults in an example config; of course with the option of the user specifying actual names/locations) <nick_1>pegaso: mean Kernel Hardening, btw was asking about TUI package configuration, do you use dotfiles? <nick_1>Rutherther: not sure what you mean by Systemd Hardening, know about Sysd-free? can you please tell me about Guix and Chroot? btw, do you use Docker / Container ? <identity>nick_1: i don't think there's linux-hardened (or -libre equivalent) in the default channel, most people manage their personal packages / dotfiles with Guix Home <identity>probably half of your questions can be answered by the manual btw <Rutherther>nick_1: sorry your questions are too abstract for me, so I have no idea what kind of answer you expect, ie. 'tell me about guix and chroot' I have no idea what to tell you about those topics. Guix just runs on linux so stuff like chrot just ports to it the same way as other distros... so I don't know what would be there about 'Guix and chroot'... <nick_1>Rutherther: some distros offer custom install etc, asked about chroot so you can execute packages there, unless you refer Docker / Container <nick_1>identity: AFAIK some Distros contain hardening by default <nick_1>know about different Kernel versions ? <emacsomancer>that's where you might look at e.g. CachyOS's hardened kernel, no? <nick_1>Rutherther: do you use Security Enhanced? <Rutherther>ieure: are your partitions still not getting unmounted cleanly after updating to guix with shepherd 1.0.3? <nick_1>ams: what packages you use ofen ? <ieure>Rutherther, One is, another is not. <nick_1>pegaso: you use dotfiles for packages ? <Rutherther>ieure: one partition? which one is unmounted cleanly? <ieure>Rutherther, Two partitions, /boot/esp with Grub, and a LVM container partition with one volume holding the root FS. ext4 for /; LUKS1 for the disk encryption. <ieure>Rutherther, Both systems have the same FS setup. They were created by the graphical installer. <Rutherther>ieure: so you meant that one system is fine and another is not? <ieure>I have one laptop that still exhibits the problem and one which no longer does. <Rutherther>can you send the config of the one that is problematic then please? <ieure>Rutherther, I can, but it depends on both my personal channel and nonguix. I got tired of repeating the same stuff for the Several machines I run, so shoved as much as I could into helper functions. <Rutherther>also, do you get this behavior every time, even when you just boot and shutdown, or only sometimes, and if sometimes, do you have a guess on what triggers it? <Rutherther>ieure: I don't mind it containing personal channel, as long as the channel is public it's fine for me to run in a vm <ieure>Rutherther, I think it's still happening on every boot, but will have to double check. <ieure>Gotta wrap up some work stuff, then will experiment / send config. <Rutherther>thanks. I won't get to it today anyway, going to sleep soon <ieure>Rutherther, Want it via email; in the bug report; mention here; something else? <Rutherther>bjoli is probably experiencing it (got corrupted file in system gen parameters file) as well and they are on guix just a few days old <identity>nick_1: lot's of questions! most of them can be answered by Just Trying It Already! <nick_1>can you please tell what Coreutils tools you use often? btw you use busybox? <ieure>nick_1, Honestly struggling to understand what you hope to learn from this question. <ieure>Like, everyone to raise their hands and say, ls, rm, mv, cp? <identity>nick_1: ls and cat. i'm not sure what that question is trying to achieve. i don't use busybox because guix comes with GNU coreutils and i just use them. <bjoli>Rutherther: this is my system config. Nothing fancy. The commented out file system is the one that made my system not boot because I accidentally did got the uuid wrong: https://paste.debian.net/1365735/ <bjoli>Sorry about the latency of replies. Life gets in between me and my computer. <bjoli>I will hopefully have some time tomorrow as well. <Rutherther>bjoli: thanks. Note that you are probably affected by a bug where the root file system is not unmounted properly, you can try running "sync" before shutting down your computer to minimize the damage. As for the damage already done (the parameters file at least), I hope I already shared the steps to 'fix' that, if you're unsure, don't hesitate to ask <vagrantc>i know there are a few language-specific stuff (haskell? julia? ada?) that are rooted in their own binary seeds ... is there a list anywhere? <Rutherther>"applying 117 grafts for kwin-6.1.4", a lot of grafts <Rutherther>oh, "applying 127 grafts for plasma-desktop-6.1.4 ..." <mirai>can someone help me troubleshoot this grub rescue trouble? <mirai>its complaining that it can't find the btrfs system although grub doesn't even try to decrypt the drive, it just goes straight into rescue mode <Rutherther>mirai: how does the generated grub.cfg look like? <ss2>specifically, what do the first three lines say? <ss2>I have /boot on an unencrypted partition. I haven't set up a fully encrypted system yet. <Rutherther>then I have no idea, you can try executing the cryptomount manually if you will get an error or something, but if not, I am out of ideas <ngz>It "@:" Texinfo markup recognized in package descriptions? <nick_1>pegaso: what Developer tools you use often? merge / diff ?