IRC channel logs
2026-01-19.log
back to list of logs
<ieure>optimal, Your key needs to be in the keyring branch, both in the remote and your local clone. Check it out and/or pull it. <optimal>ah, so i need to add my key to the keyring before i open any pull requests? <PotentialUser-72>does anyone get the issue of Flatpak apps not appearing in menu (in this case on KDE) <ieure>optimal, Your key needs to be added by an existing committer before you can commit. <ieure>optimal, This should have happened when you were given commit access. <optimal>i'm not trying to commit to the Guix repository and i don't have commit access, i'm trying to commmit to a fork <ieure>optimal, I see. Simplest thing is to disable the Git pre-push hook. <ieure>There's probably a CLI option to ignore them. <optimal>ieure, i see, thank you. someone ought to mention that in the manual's Contributing section <optimal>the option seems to be -h, or --no-verify <ieure>optimal, The manual covers contributing to Guix, so that's not really relevant if you're not doing that. <optimal>ieure, i am trying to contribute to guix though, i'm pushing to a fork so that i can open a pull request, i'm confused am i doing this incorrectly? <noe>optimal, I think its a malfunction of the hook <optimal>noe: yeah, it should probably only be enabled for commiters <noe>If you look at it (in .git/hooks/pre-push), there is a check that disables it if you are not pushing to upstream <ieure>optimal, By "a fork," do you mean "your fork?" Noting that you can use AGit to contribute without needing your own fork. <noe>optimal, did you run guix git authenticate manually before building guix? <noe>that could have installed the wrong hook <optimal>noe: yeah it seems the wrong hook is installed <optimal>there's a comment that says that guix git authenticate installed it :/ <noe>makes sense, I think you can safely delete it and it will be replaced <bigbookofbug>does the current build of qtile on guix support the wayland output yet ? i was looking at it in wm.scm and i see wayland dependencies present, but it doesnt seem like pywayland is, nor any wayland sessions defined for it <PotentialUser-22>does anyone have the solution to the problem of correctly setup fprintd not showing option to unlock plasma lockscreen wit fingerpritn <noe>bigbookofbug, “qtile start -b wayland” says that the wayland backend is not built <noe>As you point out its possible that some dependencies are missing for that, I suppose you can try adding them and see if it includes it in the build <bigbookofbug>i believe id attempted sometime in the past but i was still pretty new to guix at the time so the patch was shoddy <bigbookofbug>iirc there were some issues with python versioning as well, but i might see if that's the case still and if not submit them <noe>Don’t hesitate to post on help-guix if you hit a blocker and I’ll try to help further :) <bigbookofbug>is this a channel ? i don't believe i've heard of help-guix <oliverD>Any ideas on how I would get gnome working faster or start Icedove? <RavenJoad>I have a beefy computer where I tend to build many packages at once (It feels like I am currently rebuilding most of ghc-*). Can I bump max-jobs and reduce cores in guix-configuration? Is the only way with the extra-options list? I just want to make sure before I start doing something stupid. <RavenJoad>Most packages don't benefit from -j128, so I want to build more things in parallel with less parallelism per-job. <ieure>oliverD, What do you mean "get Gnome working faster?" Does it run slow for you? <acidbong>RavenJoad: you can adjust build users and jobs per build with --cores and --max-jobs respectively <RavenJoad>acidbong: I generally want that behavior globally on this system, so I modified guix-configuration in my system config instead. <csantosb>Hi Guix ! Correct me if I'm wrong, I understand that rebasing a pr triggers @guix-cuirass-bot, right ? Ref #5489 <identity>test202020: the manual says «Guix System is [not] available on […] ‘riscv64-linux’.» so i guess nowhere <efraim>are you looking for the binary install tarball or a system image? <PotentialUser85>If I remember correctly, guix uses guile for it's Scheme implementation. Guile does support other Programming Languages, and Lua is being worked on. While I do think Lua is easier for Beginners, this is mostly personal: Will there be/Is there any interest in making guix usable with Lua? <identity>PotentialUser85: we should start with making Guile's other language implementations usable <PotentialUser85>I've only tried the ECMA Script one very briefly, but from that small adventure I agree <Franciman>what does it mean for guile to support other programming languages? I thought it was a scheme implementation <identity>Franciman: you can lower a language to one of Guile's intermediate representations, which ends up running on the Guile VM <PotentialUser85>`guile --language=ecmascript` should put you in a repl with ecmascript. I don't know how it works from there though <identity>that is also what the work for Guilemacs (Emacs running on Guile) is based on <efraim>I realized on my powerpc machine I was running the daemon from a guix checkout almost 4 years old, built using guile-3.0.2. I've pulled and now rebuilding the checkout and hopefully that'll clearup the strange errors I keep getting there <Kabouik>I installed Guix on a foreign distribution (Debian) for a student and for some reason, packages installed from Guix don't show up in her DE. I did the same a few days before on another machine and had no such issue. I think I already added the required lines to load profiles in ~/.bash_profile: https://0x0.st/PKWB.txt <Kabouik>I can't think of anything done differently on the two PCs, not sure what is wrong. <csantosb>You should setup the env variables with /etc/profile.d/zzz-guix.sh, nothing goes in ~/.bash_profile <Kabouik>But /etc/profile.d/zzz-guix.sh was there by default, am I supposed to alter it? <csantosb>I meant, at boot, you already source all /etc/profile.d contents, including zzz-guix.sh <jas>i see the same -- i only see guix tools in the terminal on debian. i'm not sure /etc/profile.d/ is sourced at all in a desktop login, at least not until a terminal is started... <jas>whats the two foreign distros? i think /etc/profile.d/ files used to be sourced by older X11 logins, but in a modern trixie wayland setup, I'm not sure how to make that happen <yelninei>Kabouik: the environment variable you are looking for should be XDG_DATA_DIRS <yelninei>but it might be that this is not getting set to include the ~/.guix-profile/share automatically <Kabouik>It's a Debian distro, running with KDE Plasma (x11) so far as she chose that DE (but it's the same on both machines). <jas>okay, my experience is with gnome on a debian trixie derived OS, it may be that different DE's behave different here... <Kabouik>Maybe one of the only differences between both machines is one had Gnome installed during the initial setup, while the other only had KDE <Kabouik>Maybe KDE didn't install the required XDG packages <Kabouik>Or maybe Gnome came with some profile configurations that would do the required work on the necessary ENV variables <csantosb>But this is probably distribution dependent <test202020>efraim: somethink what i can install on my risc-v machine <PotentialUser-9>Hello. I was on here last week as PotentialUser-37, and user Rutherther kindly informed me that the xe-guest-utilities-service-type does add files necessary for shepherd to start the daemon as a shepherd service. I was under that impression, but have no idea where these files are or how to use them to set the daemon to start on boot. <PotentialUser-9>So my questions are two I guess: 1) how to you find out what system services are available/installed, 2) how do you set one of them to star on boot? <identity>PotentialUser-9: did you not ‹guix pull› after installing 1.4? it is seriously outdated <identity>if you did pull, then you are no longer on 1.4 <PotentialUser-9>ah okay. I installed 1.4 from the install iso, didn't realise it had done a version update <noe>use the guix describe command <identity>you can find the available services in the manual, e.g. (info "(guix) Virtualization Services") describes the ‘xe-guest-utilities-service-type’ <noe>PotentialUser-9, you can see running shepherd services with herd status <PotentialUser-9>identity I did find out about the ‘xe-guest-utilities-service-type’. But I am unsure what to do with it. The services part of the manual is not clear to me. <PotentialUser-9>I found where other services are listed in my config.scm, but adding xe-guest-utilities-service-type didn't work. I assume more configuration is required. <identity>PotentialUser-9: you add them to the ‘service’s field of your system configuration, e.g. (services (cons (service xe-guest-utilities-service-type) %base-services)) <PotentialUser-9>Unfortunately I am away from the system today, and cannot recall the error. I will have to follow up tomorrow. Thanks for your help. I believe I am on the right track. Better than feeling totally lost! <yelninei>hmm: Pre-boot error; key: misc-error, args: ("load-thunk-from-memory" "not an ELF file" () #f)Aborted <yelninei>Rutherther: Trying to build the guix package with %bootstrap-guile as the driver to see if that helps with the "repeated allocations of large block" warnings and eventual oom on i586-gnu <Rutherther>yelninei: I see. Btw is the guix pull issue present onky on i586, or also x86_64? <yelninei>on x86_64-gnu it hangs instead. Is taht better than segfaulting? <Rutherther>yelninei: well the outcome is the same, pull doesn't work :) <yelninei>But x86_64 is very far away from building the dependencies as many tests hang. i586 is only 4 patches away <RavenJoad>Is there someone from the Haskell team here? I am updating QuickCheck (and seemingly the entire Haskell universe) and am getting a multiple top-level definitions error for auto-generated properties, e.g. "Multiple declarations of ‘prop_DataBitsIff’". <yelninei>Rutherther: Even then (guix self) cannot build itself (no guile-avahi, guile-launcher needs to link with libgc) <mthl>begin; update assets set artist_uid = null where artist_uid is not null; <Rutherther>Hopefully, you never know. You would also think random fields on the web wont be sql and sometimes they turn out to be :) <podiki>anyone come across the error (on system build) "guix system: error: no target of type 'special-files' for service 'syncthing'" ? <podiki>no changes to syncthing in config or in guix, no idea what this is coming from <Rutherther>podiki: hm, are you using %base-services that brings in special-files service? I mean it shouldn't matter if you are or aren't, but if you were, it would make it even more strange <Rutherther>podiki: in any case, can you try doing guix repl on your config? <podiki>i'm trying to track down a weird thing (bug?) of a referenced /etc/guix/channels.scm that has the old savannah url <podiki>so i probably accidentially removed base-services.... <Rutherther>hm, but still...typically when a service A extends B, B is brought in automatically <podiki>that was it; dropped off %base-services completely in services list <Rutherther>I will have to re-read fold-services I guess, because this seems weird to me <podiki>i'm never sure about service requirements, like in home services pipewire needs you to explicitly add dbus <Rutherther>but that's something else, that's a shepherd requirement from a shepherd service <podiki>never tried a system without base-services <Rutherther>ah I see it now. It's that special-files does not have a default value. That's why it cannot be automatically instantiated with the extension. <Rutherther>not sure why it doesn't have a default-value though, seems perfectly fine for me to have '() as default value, for this use case <Rutherther>the procedure instantiate-missing-services that is responsible for this does not instantiate when there is no default value <podiki>yeah i don't see why you wouldn't want default values for services in general <Rutherther>I think it makes sense in some cases. Sometimes you just have to choose something as a user. But here it does not seem to be the case to me <podiki>still this mysterious channels file persists, can't gc it as it is referenced somehow <Rutherther>like sure, you wouldn't want to instantiate "(service special-files-service-type)" by itself, but who would do that, and if they did, it does not matter much <podiki>right, or warn if there is no effect <yelninei>with %bootstrap-guile there are only the warnings when translating the manuals (and then fails in the guix tests). I dont understand how changing the build driver affects anything when all it does is (invoke "make") <Rutherther>hey, could someone restart the ci.guix.gnu.org cuirass frontpage? <hanker>Hello, I have an Emacs problem, related to Tramp. The long-short story is that in `emacs-org`, `ob-eval.el` has an `org-babel--get-shell-file-name` that's used to provision the shell used in SRC blocks evaluations. This function relies on an hardcoded path, usually "/bin/sh". but replaced with the /gnu/store- path in Guix's case. However, this function can be used across Tramp, so when using it across non-Gui <hanker>x distributions, the function tries to access the file and fails. How can this be solved? <acidbong>currently building Guix locally (for the sake of contributing) with -j12, but it still compiles stuff in -j1, is that correct? <Rutherther>if you mean the guix package, yes, unfortunately, it is correct <Rutherther>guile packages aren't reproducible with parallel builds so the guix package builds with just one job <FuncProgLinux>I got a review on a MATE PR but the ACLOCAL_PATH phase I placed on the package definition was taken from a GNOME package I think. The issue is that I do not understand the instructions or what am I doing wrong there ._. <yelninei>janneke: building guix With guile@2.0 as the driver is not leaking anything. Maybe the change to use posix_spawn for system* has a leak somewhere? <nckx>Rutherther: Done, looks OK now. Thanks. <mthl>I have a pull request containing fixes and improvements for the clojure packages <https://codeberg.org/guix/guix/pulls/5162>. I submitted it 3 weeks ago and it starts feeling a bit long. Is there anything that I can do to help making it move to master or next-master ? <noe>mthl, I’ve added a comment. A good thing to do would be marking the finished conversations as resolved (“Resolve conversation” button next to reply) and then you can ping the reviewer for a new review <mthl>noe: Thanks for your advice, I will do that. <noe>FuncProgLinux, maybe try to find why ACLOCAL_PATH is not taken into account <FuncProgLinux>noe: I think it was part of the autogen.sh script, I remember I tried to substitute* my way but didn't work and several days into my first attempt I read a package definition under gnu/packages that had this specific snippet and just pasted it. <noe>oh right I can find it in libgpod <FuncProgLinux>noe: AFAIK what it does is -I /path1 -I /path2 -I /pathX isn't it? <FuncProgLinux>On top of that I think I hit a reproducibility issue with that package because sometimes it builds, other times itstool errors out in merging translations with TypeError: 'NoneType' object is not subscriptable <psycotica0>I have the same laptop that I use at an office I rent and at home. And that same device can run `guix pull` at home, but not at the office. <psycotica0>But the office WiFi isn't locked down as far as I know. And I ssh'd into my house just now to make sure it's still working there, just to make sure it's not an outage or something. <fezbpu>Hey all, I'm trying to install emacs via guix under Ubuntu 24.04 LTS on WSL (convoluted I know), and i can't get it to install a version newer than Emacs 28.2. `guix pull` refers to commit c0c184a, which is the latest commit from an hour ago. <mthl>I have a question regarding the return value of custom build phases. What I have in mind is that they were supposed to always return a boolean value. Does this rule still apply nowadays or is returning void fine ? <sleepydog>does shepherd have readiness notification? I have a race between when a service creates a unix socket and when a dependent service tries to open it.