IRC channel logs
2024-10-29.log
back to list of logs
<apteryx>hm, the whole network is down? hydra-guix-129 is also out of the blue <bufu112>hi all, hope it's okay to ask for help here, was really excited to jump into guix today after work after using nix for a little while, my installation of guixSD is failing out consistently right when it gets to trying to build/load xz? ends with guix system: error: corrupt input while restoring archive from #<closed: file 7fb80780c4d0>, and then <bufu112>kills with signal 7 :( can't find much online because it seems like the boards are down! <apteryx>I guess you are lacking substitutes due to one of the build farms being unreachable at the moment <apteryx>the corruct input error message is odd though; either the connection is failing or the archive you got is really corrupted <bufu112>ah, that makes sense! would i only be able to run that command from a manual install? i wouldn't mind going through that way, but i was just doing graphical, and didn't see a way i could slide that in. <bufu112>in the mean time, yeah, i'm wondering if my file is just corrupted... i reflashed my drive and redownloaded the iso and am attempting again, so fingers crossed it's just my own fault for not checking the hash :P <apteryx>this is an option for the guix-daemon, so if you were to 'pgrep -a guix-daemon' to get its current command line, then herd stop guix-daemon to stop it, then run it manually with the right --substitute-urls option, that should allow you to continue with the graphical installer <marmar>flatpak apps take at least 5 minutes to start up. Anyone know how to troubleshoot this? <marmar>I suspect it's something related to that socket error, but I don't know where to go from there <podiki>yes and others hosted on the berlin servers like ci.guix.gnu.org <podiki>known but i don't know the reason or status personally <nutcase>marmar: I have the same issue. I read that removing .local/share/flatpak and reinstalling the flatpaks should solve the issue at least temporarily. For me this didn't solve the issues. <nutcase>marmar: I also read that it could be related to (not) running the wrong xdg-desktop-portal <adanska>oop, is everything okay with guix ci? <marmar>nutcase: oh I tried removing .local/share/flatpaks months ago with no success <marmar>nutcase: I'm currently using plasma6 wayland, so I don't know how to check on xdg-desktop-portal <Rutherther>nutcase: having both wlr and gnome at the same time is quite strange <nutcase>Rutherther: that's what I changed recently in hope for getting rid of this issue. Without gnome-portal it's the same <nutcase>ACTION removes the gnome-portal again <Rutherther>nutcase: how did you remove it? also probably better to actually check what's going on - like looking into log of xdg-destkop-portal ran with -v <nutcase>Rutherther: I removed it from my operating-system's packages, reconfigured, logged out and in <Rutherther>nutcase: this "XDP: Falling back to gnome-keyring.portal for org.freedesktop.impl.portal.Secret" is a problem <nutcase>I think, I don't have any particular configuration for the desktop-portal(s) <Rutherther>nutcase: hmm, I don't really see an issue in your portal log then <nutcase>Rutherther: anyways: thanks for checking <nutcase>Rutherther: you don't have any issues and are running the same (versions of) desktop-portals? <migasguix>Hi, i'm trying to install guix in qemu to start learning it, but i keep running this error in install <nutcase>Rutherther: desktop-portal 1.16 is outdated and 1.18.4 is available but not the version I'm running <PotentialUser-4>Hello. I installed package "tlp", but tlp refuses to run because there is no /etc/tlp.conf. Is this normal or expected for a Guix package? <Rutherther>nutcase: I don't use flatpak, but no problems, I have wlr and gtk portals, latest versions in guix channel <Rutherther>migasguix: unfortunately it seems that both substitutes are currently off :( you can use --no-substitutes flag to get around it, but I suppose you won't like to wait for everything to build, especially in a vm :D So maybe just start later... <Rutherther>PotentialUser-4: some packages need configuration, guix system doesn't work like other distros, where you get the config as part of the package. Only installing it means its only put to a specific profile. You typically use services to install packages including configuration. See tlp-service-type, you can just use that one <PotentialUser-4>Oh, I see. The service would then install tlp as a dependency, right? <Rutherther>PotentialUser-4: it extends other services, mainly udev and shepherd in this case. You won't even have tlp executable in your path, but there will be a service running <PotentialUser-4>I see. Thus, if I wish to interact with the tlp service I should keep the package installed. <jakef>is ci.guix.gnu.org down for anyone else? <PotentialUser-4>When I took a look at the patch-tracking database I noticed quite the backlog. Some patches had been submitted to support a newer version of the software and appeared ignored (no comment on them, but also not accepted). Why would that happen? <nodata>hi, when doing `guix pull`, with guix installed on foreign distro, I get multiple "warning: 'postgresql-11' is deprecated". I did not install postgresql explicitly, in fact there is almost nothing installed there. Does anyone know how to handle this message? <Guest5>Hi people. I'm having a problem, and I can't find a solution to it. It's related to GTK icons. Guix GTK applications seems loading correctly GTK themes defined in settings.ini file, but they aren't loading GTK icon themes defined in `gtk-icon-theme-name` <Guest5>It's weird because, apt installed applications seems to load correctly any icon theme, but Guix installed applications not. <lilyp>are you missing librsvg by chance? <Guest5>Seems like is missing. I tried `guix package --list-installed | grep librsvg` and got nothing. Let me try installing it- <PotentialUser-4>nodata I got that message also on Guix System. I thought it would mean that PostgreSQL 11 was going to be removed, but version 10 is still in the repository. <lilyp>Both substitute servers are having troubles right now – you might want to wait a little <cricri>Hi there, I get a 'hash mismatch for store item' for the package libgcrypt when I try to build an image with 'guix system image ...'. Is there something I can do about it? I have alreaded tried a 'guix pull' with no success. Thanks! <Guest5>cricri: It seems like the servers are down <Guest5>Nor `wget` works. It keeps saying `Host is unreachable` <cricri>thanks, but how does it relate to the hash mismatch? <Deltafire>cricri: i had this with the signal package a few weeks ago, i had to force it to build the package instead of using the presumably corrupt version from the build server <cricri>so passing --no-substitutes to the 'guix system image' command might help? <Guest5>cricri: No. Because substitute servers are down too <Rutherther>cricri: probably the file changed upstream. It's not common, but it happens, ufortunately. Will need to be investigated and probably updated in guix. For now you would have to change the package yourself <Guest5>The only thing we can do is wait, ig <ennoausberlin>Hi Guix. I follow the discussion about financing on the mailing list. A payment for trusted infrastructure maintainers would be nice. I often run into infra outtages - as of now - while showing guix to others. The first impression is often not that good, then. Thank you for the volunteer work, anyway. <cricri>Guest5: Thanks a lot! It is trying at the moment... <graywolf>Hello :) Would anyone have some pointers how to mount 9p file system in the operating-system definition? I cannot figure it out :/ <graywolf>In particular, I cannot figure out what to put into (device) field. <Rutherther>cricri: I reported the problem with libgcrypt, it's tracked as #74089 <Rutherther>ouch, did peanuts just leave because it could not contact issues.guix.gnu.org? <ArneBab>Is anyone working on updating icecat? I tried to simply build the patches against FF 128, but that failed. Is someone already working on updating the patches for FF 128? <cricri>Rutherter: thank you, there are no complaints about the hash anymore :) <Rutherther>cricri: so you don't get it when you do "guix build libgcrypt --check --source"? <futurile>ArneBab: don't think I've seen anyone working on it <Guest5>Hello. I'm from the gtk icon question. Now everything got worse. I installed `librsvg` via `guix install`, but now my pcman (installed via my native package manager) don't run anymore <Guest5>As well other applications are not working well now. <Guest5>It seems like the application is trying to load the Guix installed libraries when it should load the native libraries <Guest5>When I run `LD_LIBRARY_PATH=/usr/lib pcmanfm` it works. So that means... Guix overrided the default `LD_LIBRARY_PATH` value? <Guest5>How can I do my guix applications use its recommended `LD_LIBRARY_PATH` without make something horrible like modify `.desktop` entries? <Guest5>Because Guix applications works well now, but I don't like this `LD_LIBRARY_PATH` problem in my system. <cricri>Rutherther: let me try again, when this build is finished. I am doing it in an aarch64 vm on a x86_64 machine, so it takes a while... <futurile>Guest5: it's because when you source the ~/.guile-profile/etc/profile it setting things like XDG_CONFIG_DIR. If you don't want to use Guix's "graphical" apps remove them and it will top trying to set XDG_* <Guest5>futurile: Before installing `librsvg` with `guix install` my system applications worked well, but guix graphical applications had problems on displaying GTK icon themes <Rutherther>cricri: the command I shared only downloads the source, it doesn't build anything else <PotentialUser-87>Hi sorry for just popping in, but I can't seem to reach ci.guix.gnu.org. Is it down for service right now? <Guest5>And now, if I remove `librsvg` with `guix install` I'll get my system applications working well again, but (again) guix graphical application will not load icons anymore <Guest5>PotentialUser-87: Try using the other mirror <Rutherther>Guest5: so if you install a package with guix it doesn't work, but if you have _the same_ package with system package manager it does work fine? <Guest5>PotentialUser-87: You're welcome :3 <Guest5>Rutherther: Guix graphical applications works, but displays incorrectly icons on GTK applications. This because the `librsvg` package is missing for Guix (even whe I have it installed natively on my system. I'm using Alpine btw) <Rutherther>Guest5: the same applications or different ones? <Guest5>Rutherther: I didn't understand pwp <fnat>We don't have a very well defined agenda today, just catching up and potentially some patch review. <Guest5>I have `librsvg` installed on my system (without guix), and all my graphical applications _installed with the system native package manager_ displays correctly GTK icons. But Guix applications don't <Rutherther>Guest5: have you tried running the same gui application with both system and guix, where the icons were working only in the one installed from system package manager? or are they only different applications? <Guest5>When I install `librsvg` occurs the inverse case. All Guix application works well and display GTK icons correctly, but native applications don't work well anymore, because `LD_LIBRARY_PATH` is trying to load the Guix `librsvg` stuff <Rutherther>Guest5: so what sets your LD_LIBRARY_PATH, why do you have it set? <Guest5>Rutherther: 1) I haven't tried that. Maybe it would work, but I want to run `apk` installed applications too. 2) I haven't set it \:( But all my applications are trying to load guix libraries anyway <Rutherther>Guest5: I am not asking so that you would do it in the end. I asked because I wanted to rule out a different reason than the packages being from Guix. Not all packages behave the same when it comes to this, even gtk3 and gtk4 can differ. But you've already established when installing librsvg it starts working, so you don't have to try it anymore <Rutherther>Guest5: so what is your value of "LD_LIBRARY_PATH"? <vhns>Hello. I currently have a (modify-services %desktop-services (delete gdm-service-type)) clause within my (operating-system ... ) but I still get gdm on startup. Should I be using some different name? <Guest5>Rutherther: 1) All GTK applications had this problem until I installed `librsvg` (with `guix install`). So these applications didn't located my native `/usr/lib/librsvg-*.so.*` 2) There's nothing (literally `echo $LD_LIBRARY_PATH` displays nothing) <Rutherther>vhns: do you by any chance also have (set-xorg-configuration in your config?v <vhns>Rutherther: I do '(set-xorg-configuration (xorg-configuration (keyboard-layout keyboard-layout))) <Rutherther>vhns: if you are using startx command service type, then just pass the xorg configuration to that instead <vhns>I'm migrating off of xfce4 to sway(wayland) and forgot to remove some xorg stuff from my os config file <Rutherther>Guest5: so why would the system applications try to load librsvg from guix when there is nothing in ld lib path? that doesn't check out to me. Could you send the error you get when you have librsvg installed via guix? <Guest5>(pcmanfm:27139): Gtk-WARNING **: 15:14:45.920: Could not load a pixbuf from icon theme. <Guest5>This may indicate that pixbuf loaders or the mime database could not be found. <Guest5>(pcmanfm:27139): Gtk-WARNING **: 15:14:45.920: Error loading theme icon 'document-new' for stock: Failed to load /home/youw/.local/share/icons/Zafiro-Icons-Dark/actions/16-Dark/document-new.svg: Unable to load image-loading module: /gnu/store/hsh3wr9kr99yj2q8v3vhy1lca6h4g0c8-librsvg-2.56.4/lib/gdk-pixbuf-2.0/2.10.0/loaders/libpixbufloader-svg.so: <Guest5>Error relocating /gnu/store/zzpbp6rr43smwxzvzd4qd317z5j7qblj-gcc-11.4.0-lib/lib/libgcc_s.so.1: __cpu_indicator_init: symbol not found <Guest5>(pcmanfm:27139): Gtk-WARNING **: 15:14:45.922: Error loading theme icon 'image-missing' for stock: Failed to load /home/youw/.local/share/icons/Zafiro-Icons-Dark/actions/16-Dark/image-missing.svg: Unable to load image-loading module: /gnu/store/hsh3wr9kr99yj2q8v3vhy1lca6h4g0c8-librsvg-2.56.4/lib/gdk-pixbuf-2.0/2.10.0/loaders/libpixbufloader-svg.so: <Guest5>Error relocating /gnu/store/zzpbp6rr43smwxzvzd4qd317z5j7qblj-gcc-11.4.0-lib/lib/libgcc_s.so.1: __cpu_indicator_init: symbol not found <Guest5>(pcmanfm:27139): Gdk-CRITICAL **: 15:14:45.922: gdk_cairo_surface_create_from_pixbuf: assertion 'GDK_IS_PIXBUF (pixbuf)' failed <Rutherther>Guest5: please send them with paste site, not as messages. irc is a line based protocol, every line is a message <Guest5>Rutherther: Sorry, my bad practices- <Guest5>This output is by running `pcmanfm` without setting `LD_LIBRARY_PATH` (so it tries to load the Guix `librsvg` library) <Rutherther>Guest5: so it tries to load the Guix "librsvg" library - when ld library path is not set, it shouldn't be doing that, it should be loading only from default paths <Rutherther>Guest5: if you explicitely do "LD_LIBRARY_PATH= pcmanfm", you still get the errors? <Guest5>When I run it with `LD_LIBRARY_PATH=/lib:/usr/lib pcmanfm` everything works fine <Rutherther>Guest5: that's not what I meant, I meant with empty ld library path <Guest5>Rutherther: Now I tried it (`LD_LIBRARY_PATH= pcmanfm`) and outputs the same as I show you on the pastebin <Rutherther>Guest5: okay, it seems I don't really understand what the gtk apps are doing. :( could you try starting guix gtk app with "LD_LIBRARY_PATH=/gnu/store/hsh3wr9kr99yj2q8v3vhy1lca6h4g0c8-librsvg-2.56.4/lib", without having librsvg installed? if it worked, one solution would be to wrap all the packages and install the wrapped variants. Not pretty, but it would solve all the issues I think <Guest5>That makes me think one thing... Maybe all this time all my programs was loading Guix libraries instead of native (`/usr/lib` and `/lib`) libraries. What a mess... <aurtzy>Guest5: i just checked some old foreign distro configurations of mine and i have a note about GDK_PIXBUF_MODULE_FILE causing issues; maybe unsetting this works? <Rutherther>Guest5: I doubt that all of them. This seems like some gtk magic. The dynamically linked executables should definitely not be just looking to your ~/.guix-profile/bin, unless you had it in ld library path <sepeth> For those who are interested, there's a patch session happening in guix-london-meetup group, you can join here: meet.jit.si/london-guix-meetup (sorry if this was announced earlier, I couldn't see and thought some of you might be interested). <Guest5>Rutherther: Running `LD_LIBRARY_PATH=/gnu/store/<long-id>-librsvg-2.56.4/lib pcmanfm` (obviously) doesn't work, but `LD_LIBRARY_PATH=/gnu/store/<long-id>-librsvg-2.56.4/lib remmina` (for example) works as expected <Rutherther>Guest5: yeah, that's what I meant, to try running guix apps with that, not the system packaged ones. You could potentially wrap the guix apps to run with that ld lib path then. It would involve some code in guile. Not sure about a better solution :( <Rutherther>upon checking guix source it seems that some packages are wrapped with GDK_PIXBUF_MODULE_FILE, not sure why remmina isn't, maybe it should be <Guest5>aurtzy: Unsetting this variables causes `remmina` don't display correctly GTK icons (again) <Rutherther>Guest5: yeah, you would need to set it for guix apps only. That means wrapping them <Guest5>Btw guys, what should I do if I want to install the GNU Guix operating system on my machine but I need a (horrible) propietary firmware for my Realtek network card? <vagrantc>Guest5: there are probably other repositories on the internet that allow such things. though it is awkwardly not within policy to directly reference them in guix channels. <vagrantc>due to guix following the Free Software Distribution Guidelines... <[>or, you could get a better network card <krascovict>Hello everyone, I was analyzing the server debug graph and I detected a pattern... increasing the system tables does not affect the cache usage, or merging them using Schema for that? <Guest5>vagrantc [: I'm saving money to get a T440p as fast as I can. But for now I need to do a "deal with the devil" <krascovict>I'm new to GUIX, I thought it would be interesting to open a debate to optimize the server's own processes. <krascovict>I had an idea to create a "swap cache" in Scheme and do the optimization at runtime without depending on third-party binaries. <vhns>Does Guix support pipewire? <vhns>nvm, I'm stupid, should've read the docs better <Guest5>vhns: I guess it do. Did you tried `.guix-profile/libexec/pipewire-launcher`? <vhns>Guest5: There is a section dedicated to pipewire under "(guix)Sound Home Services" -- PipeWire, home service <Rutherther>Guest5: usually better to run stuff like that via the home services, but manually launching could also work, yeah <Guest5>Rutherther: Yeah, I think so. Sorry, I'm not using GNU/Guix operating system :( <Rutherther>Guest5: you can use guix home even on other operating systems <dariqq>Hi, I was playing around starting my home shepherd via the gnome autostart instead of the default autostart to have things like WAYLAND_DISPLAY available in the emacs-daemon <dariqq>Problem is adding home-dbus-service makes gdm think the user dbus is already running (i guess via the env var) and refusing to log in because it cant reach dbus <podiki>i have home-dbus and use gdm (though not to run gnome) but haven't had that issue <dariqq>podiki: Do you use the guix autostart way to launch the home shepherd? because that works with home-dbus but forks off to early to have WAYLAND_DISPLAY in its environment. The gnome x session is fine for some reason <podiki>if you mean that guix home launches a shepherd then yes; also don't know if i have anything from there (i only use a few little services and need to move more there) that needs wayland_display <dariqq>apologies it is gnome-session-binary and not gdm that is complaining <Rutherther>dariqq: I suppose you won't have a great time with forking off like that, since your session won't get the same dbus socket as what guix home starts services with. Or did you set the variable yourself somewhere and now the problem is that it's not yet running even though the variable is set? <dariqq>Rutherther: the environment gets set via home-dbus-environment-variables and is set upon login i think. And in this case it it is problematic because I wanted to launch shepherd later to have my proper environment there. but because home-dbus is not yet actually running gnome-session is confused <dariqq>and by confused i mean there is an error and gdm brings me back to the login menu <dariqq>maybe there is a better way to update the environemnt of shepherd and all services? <vagrantc>now jsut need to convince the debian security team to approve fixes too ... <Rutherther>vagrantc: what version of guix did it use before? <vagrantc>Rutherther: only using released versions, so 1.4.0 ... although a very old version uses 1.3.x <dariqq>Rutherther: I guess that works, but would like something that works independantly of the session type <Rutherther>dariqq: yes, you just have to start the service(s) somewhere, like with the autostart you have, but instead of starting shepherd you start the services <dariqq>yes and I would need a emacs that depends on wayland display, and a one on x-display for when i am in an xsession (however that works without issues with gdm)? <Rutherther>dariqq: no, just get both variables, and one of them will be empty <vagrantc>hah, correction ... debian oldstable has guix 1.2.x <lispmacs[work]>Hi, I'm trying to use debbugs-gnu-guix-search in Emacs but don't seem to be getting any search results <lispmacs[work]>I see the issues Web page is down, so I couldn't double check myself that way <Rutherther>lispmacs[work]: you can still use debbugs.gnu.org to see issues <lispmacs[work]>okay, so I see that my search term should be producing results, unless I'm misunderstanding the search term syntax for debbugs-gnu-guix-search <dariqq>Rutherther: I see now. Maybe i can just use something like you do to just feed my entire environement into shepherd <Kolev>What version of GNOME is Guix on? <dariqq>Rutherther: just thought of something. If i need to do things manually anyway why dont just start emacs via 'herd load root script' where script just sets (default-environment-variables) to (environ) again and then starts emacs? <krascovict>compilando /gnu/store/w97m8cyh1mhk5zw4snphznax7qbaqsgx-module-import-compiled.drv... <krascovict>compilando /gnu/store/1lpfi9b30ljpg5pg47xcg2qsrjxg1m15-compute-guix-derivation.drv... <krascovict>Computing Guix derivation for 'x86_64-linux'... \warning: 'postgresql-11' is deprecated <wakyct>hello, I'm going to reconfigure soon and add the pipewire home service. Does anyone know if I need to do anything in system config,like modify Pulse in %desktop-services perhaps? <wakyct>the home service doc says I can enable connection by Pulse clients (the default I think) but I don't know if I need extra config 
