IRC channel logs

<civodul>i guess one rarely refers to glibc from the command line

<gabber>or their home profile - apparently

<janneke>yeah..but i believe that i do guix shell glibc sometimes (for ldd)

<janneke>ACTION notices the warning...probably i just ignore that when i see it

<janneke>civodul: do we need another guix package update for offloading to work and/or are we waiting for some other nice patches to land first?

<civodul>janneke: for offloading, you mean the issue that Tomas recently fixed in b8a45bd0473ab2ba9b96b7ef429a557ece9bf06c?

<civodul>that one needs an update of the ‘guix’ package

<civodul>no need to wait though

<civodul>feel free to do that if you have free cycles!

<janneke>ACTION goes to run make check first this time

<civodul>heh yeah, that can be annoying…

<homo>amazing update, my laptop was able to properly shutdown after system reconfigure without damaging filesystem, thank you

<janneke>hmm, guix-packages.test fails; the gnu hello package's synopsis changed

<janneke>synopsis: Hello, GNU world: An example GNU package

<janneke>synopsis: Example GNU package

<janneke>oh crap, python got merged while i'm running tests

<janneke>dude

<gabber>janneke: whoopsi, hadn't checked that when linting all of these synopses (and rewriting the one in question)

<janneke>gabber: np, i've a fix ready...but re-running all tests on the new python commits before pinning a new guix

<janneke>this kind of "janitorial" work is often overlooked and very important

<janneke>okay, the python merge breaks FAIL: tests/guix-shell-export-manifest.sh

<janneke>too bad

<janneke>ACTION goes to bed, more tomorrow -> zZzzz

<Kolev>janneke, gn! ❤️

<sneek>Welcome back weary-traveler :D

<Ironsmith>heya guix! wondering if this question should be an email, but i noticed that we reimplement a json parser in (guix build json) while guile-json is a requirement of guix and already has a json implementation. however, you can't just (use-module (json)) in build code, you have to (with-extensions (list guile-json) ...) first as mentioned in the

<Ironsmith>snippet right above this link/anchor position: https://guix.gnu.org/manual/devel/en/guix.html#index-_0023_007eexp

<Ironsmith>would it make sense to add the ability for build code to import the (json) module without having to hack it with (with-extensions ...) ?

<olafes>please, tell me there is a tool to convert Cargo.toml to a guix package

<olafes>found this one https://github.com/lelongg/carguix but it doesn't even build

<homo>olafes: check out $ info "(guix) Invoking guix import"

<homo> https://guix.gnu.org/cookbook/en/html_node/Recursive-importers.html

<olafes>yess

<olafes>homo: thank you

<janneke>tests pass, /me pushes a new guix package

<janneke>what could possibly go wrong?

<xelxebar>ACTION throws confetti atop janneke

<janneke>ACTION looks joyfully around

<janneke>ACTION locally squashes and rebases core-package-team

<janneke>and goes to see what happened after the big python merge

<sleepydog>homo: yes I noticed fontsrv in the plan9port package returns no fonts.

<sleepydog>there's some logic in the INSTALL script that I had to remove to get it to build properly

<attila_lendvai>`guix weather guix` comes back empty. when they hang the hangman... :) (or whatever is the equivalent saying in english)

<Rutherther>attila_lendvai: it's worse, it actually doesn't build, even locally

<attila_lendvai>damn! thanks for the heads up Rutherther! i was planning to rebase my pending python related patches, but i guess i'll delay that then...

<attila_lendvai>or actually, i make a pull --commit rather

<janneke>Rutherther: ow crap; do you know what is causing the failure?

<janneke>i fixed one test yesterday, then needed to rebase on the python merge, fixed another failing test

<janneke>this morning, all tests passed so i updated guix

<homo>sleepydog: "returns no fonts" do you mean it doesn't know about fonts-dir in /gnu/store?

<janneke>iwbn if all tests would pass before a branch was merged (on both branches)

<janneke>ACTION holds off on pushing a fresh core-package-team until guix builds

<sleepydog>homo: I don't know. when I'm at my computer I'll share my patch

<Rutherther> https://ci.guix.gnu.org/build/6971831/log I don't really know, since there are multiple errors, but probably some of them are just red herrings. For example I think I've seen fontconfig error for no writable cache directories even for successful builds

<janneke>Rutherther: thanks, well at least all tests pass, yay

<janneke>this isn't good: help2man: can't get `--help' info from guix copy

<janneke>weird, but that works from the work tree

<attila_lendvai>ACTION finds help2man to be mostly just a source of problems, in return for very little value...

<attila_lendvai>e.g. help2man is the only reason shepherd cannot be cross-compiled when built from git sources

<yelninei>janneke that build is guix-minimal-1.4.0-29 and not the new 1.4.0-30

<attila_lendvai>s/only/first/

<Rutherther>yelninei: yeah, -30 is currently scheduled

<Rutherther>(haven't tried to build that one locally yet)

<janneke>attila_lendvai: thanks, phew

<yelninei>right now seems like a bad time to update, some things fail to build already because of missing setuptools and lots of missing substitutes

<janneke>attila_lendvai: otoh, if 1.4.0-29 needed a rebuild after the python-team branch and it fails...well, let's wait for the actual result

<janneke>yeah, i was just a couple of minutes too late yesterday :-(

<divya>Can someone tell me why is this `guix git authenticate` failing? I checked the signature of the commits with git verify-commit: https://paste.debian.net/1339796/

<attila_lendvai>divya, guix authenticate is much more than git verify-commit. e.g. there must be a branch called `keyring` that contains the required public key

<divya>There is

<attila_lendvai>divya, and did you add your key to .guix-authroizations? and specify the first commit with your signature as the introductory commit in the channel entry?

<divya>Yep, indeed.

<attila_lendvai>yeah, some rather heavy packages are lacking substitutes; it's indeed not a good time to pull, even when leaving out the guix glitch.

<janneke>ah, /me meant to mention yelninei :)

<janneke>you probably got that already

<yelninei>janneke i dont think it would have mattered because of the test failure introduced by the merge?

<attila_lendvai>janneke, yep. re 1.4.0-29: i've pulled prior to your guix update commit, and a home reconfigure successfully fetched a binary for -29

<attila_lendvai>ACTION goes AFK

<divya>attila_lendvai: I fixed it, it was a silly parenthesis unmatch in .guix-authorizatin

<jakef>hi guix, does anyone know if shepherd timers can be used in home-shepherd-service-type?

<rekado>Hi Guix

<divya>Hello, rekado

<yelninei>oh god: gtk fails because of an error in python-pytest-cov-6.0.0

<jakef>here's info about the error i got trying to use a timer in guix home: https://0x0.st/XF1I.txt

<Rutherther>jakef: add the timers module to modules field of shepherd-service

<Rutherther>you can append %default-modules out of "guix services shepherd" to get the rest you get normally

<yelninei>ooh false alarm. I have been playing around with the toml parser and had it in my custom channel shadowing the one from guix

<yelninei>Now there are also siginificantly less rebuilds

<janneke>hehe

<jakef>Rutherther: thanks, trying to get the right syntax: is it (modules `((shepherd service timer) ,@%default-modules))

<Rutherther>that could be fine

<jakef>it's giving me "Wrong type to apply: 0"

<jsbiff>So, I installed podman using guix, and it kept giving an error that it couldn't find exe 'newuidmap' in PATH.

<jsbiff>I ended up having to install the deb package 'rootlesskit' via apt, because there apparently is no package for rootlesskit in guix.

<jsbiff>Is there a way to request a new package in guix? I'm not sure, but I think rootlesskit is free software, so could be distributed by guix?

<jsbiff>It would be good if podman would install all required dependencies so it just runs after guix install podman.

<jsbiff>correction: couldn't find exec 'newuidmap'

<Rutherther>jakef: can you send the whole thing?

<jakef>Rutherther: ok, i'll make it more minimal first

<Rutherther>I meant the error

<aemogie>hi! i noticed that keyd isnt packaged for guix, so i was wondering what the community uses for remapping keys (beyond just ctrl:nocaps)

<ekaitz>aemogie: don't be surprised if nobody needs that... we are not a huge community

<ekaitz>aemogie: you could try to package it

<aemogie>right that does make sense. but im wondering whether keyd is overkill for my usage. for something like escape on tapping capslock and control on holding it. is that possible with xkb configurations?

<Rutherther>aemogie: there are other packages, like kmonad for remapping, so I suppose people are using other packages, not that they wouldn't be remapping

<ekaitz>aemogie: idk if that's configurable with xkb... i have € some layout changing config for different languages and the € sign in the E for US layout. But i don't think you can do much more than that.

<ekaitz>s/€//

<aemogie>Rutherther: oh that looks to use a lisp to configure as well. might give that a try

<Rutherther>aemogie: yeah, and also it works on the devices directly, so works with both xorg and wayland

<aemogie>ekaitz: ah got it. im mostly clueless as to how linux's input stack works. not entirely sure what "xkb" or "evdev" or "udev" and such do and differ.

<ekaitz>aemogie: in fact, what I have is xorg config, which idk if it's the same as xkb (i have no clue)

<aemogie>Rutherther: do you use kmonad? if so could you please point me to you configuration?

<Rutherther>aemogie: I currently don't use it as I have a qmk keyboard, I will probably start using it soon for my laptop keyboard, but not yet. I was planning on using this tutorial that has an example for homerow mods, maybe it will help you, since it showcases what you want - someting on tap and something else on hold https://precondition.github.io/home-row-mods#using-home-row-mods-with-kmonad

<aemogie>Rutherther: thanks! i was more looking in the direction of guix integration. ie the kmonad documentation mentions udev rules in it's guix installation section, but i assume id also need shepherd service definitions? im going to try reading through the documentation a bit more.

<Rutherther>aemogie: ah I see. Yeah, udev rules will be useful to be able to run kmonad as user. Shepherd service could be convenient to run it, yeah

<Rutherther>aemogie: btw if you are going to add that udev service config to be able to run kmonad as user, I would recommend using "udev-rules-service" procedure that extends the service, rather than modifying the configuration of the base service

<aemogie>Rutherther: not entirely sure what udev rules even are but, using the service extension mechanisms over `modify-services` as the docs mention does make sense.

<Rutherther>udev rules are rules for devices, it controls what happens with the files that represent the devices, ie. what permissions they get, what script gets ran when they connect

<aemogie>could a udev rule also run kmonad as well (over a shepherd service)? or maybe it already does that

<Rutherther>possibly. But it's not meant for that

<rekado>I have another little Sway problem: swayidle launches swaylock after some time, but I cannot seem to unlock it again. I'm using the US English Dvorak layout everywhere. It's configured in my Guix System configuration and also in my Sway configuration. Does swaylock require additional configuration?

<rekado>(I'm guessing that this is a problem with the layout, but it could, of course, also be an obscure permissions problem)

<Rutherther>rekado: did you configure pam rules when installing swaylock?

<divya>aemogie: Hello, I also use kmonad in Guix as a service, here's my configuration: https://codeberg.org/divyaranjan/dotfiles/src/branch/master/.config/guix/system.scm

<divya>Check the `kmonad-service`. This will add it to the list of system herd services.

<rekado>Rutherther: I didn't explicitly install swaylock.

<rekado>where can I see the needed pam rules?

<Rutherther>it's a normal pam rule, called swaylock. The important is the name, just add it with screen locker service type

<Rutherther>ah, it's great that now python-toolchain is no longer the default-python, though it seems for example python-argparse-manpage is now missing: setuptools, wheel, pip dependencies. Anyone taking care of package like those or should I submit an individual patch?

<altomcat>Hey there!

<altomcat>Since my last guix-pull, the package python-libvirt failed to build.

<altomcat>I made a quick fix locally into my own channel.

<altomcat>Is there any way to run guix pull again by using my own definition package for this package ?

<altomcat>sorry It failed after a guix home reconfigure ...

<aemogie>divya: thank you!

<Rutherther>altomcat: sure, just run guix pull with url being a local guix folder

<rekado>Rutherther: thanks for the hints! It works now. I think I have enough information together to write a Cookbook entry. (Though perhaps we should consider adding a window-manager-service-type as an alternative to desktop-service-type.)

<csantosb>Hello guix !

<csantosb>If I understand it correctly, `./pre-inst-env guix build --dependents emacs-with-editor` should work, right ?

<csantosb>It should, with any package

<Rutherther>csantosb: yeah

<csantosb>Ah ! emacs-git-email doesn't build anyway

<csantosb>Which brings to b4 ...

<csantosb> https://ci.guix.gnu.org/build/7184721/details, python merge, I guess

<rekado>was the python-team branch not build on ci.guix.gnu.org?

<rekado>master is at around 50%

<altomcat>Rutherther: Thanks for the tips

<podiki>rekado: i think it was set to just x86_64 (as some other branch did recently too, don't remember which)

<civodul>rekado: did you see the reply re cloud-init? https://issues.guix.gnu.org/74389 (you were not Cc’d)

<rekado>civodul: I had not seen it before. Thanks for the pointer.

<Deltafire>why does guix do "updating substitutes" multiple times during an update?

<freakingpenguin>Deltafire: To my understanding it doesn't update all substitutes at once, just the ones it needs during that particular phase. Pretend there's a hidden number like "updating 5 substitutes..."

<Deltafire>that makes more sense.. although i still don't know how or why it has multiple phases

<Deltafire>i must have missed that section of the manual

<ajarara>hi, struggling a little: python-jupytext is causing reconfigures to fail. guix weather shows no substitutes. I found a revision where jupytext is fixed, but the real problem is I don't know what in my config is requiring python-jupytext. I'd just as soon remove it. I think it's something in %desktop-services.e

<jonsger>ajarara: assuming you are somehow using gnome: guix graph --path gnome python-jupyter-client

<ajarara>nope, just i3. Even dropping that I still end up building it.

IRC channel logs

2024-12-14.log