IRC channel logs


back to list of logs

<buenouanq>efraim: oh, yes - I forget about that. If it's been o
<buenouanq>ff for a while, I usually have to reboot twice to get normal trackpad behavior. And no native wifi, I use a ThinkPenguin USB thing.
<darkpsi>hey guys if anyone can help me there seems to be a bug with nvme the new linux kernel (4.16.1) has the nvme drivers compiled in to the kernel and when i guix reconfigure it ither demands that i put load nvme on the initramfs or if i have done that it will say it can not find nvme.ko as it does not exist. so what would be the best way to sort this problem i could ither edit the kernel config or disable the
<zybell_>There should be a file /sys/class/nvm... or similar that exist in *earlier* kernels only if nvme.ko is loaded. It would exist in 4.16.1 anyway. Extend the test to look for that file, and if lsmod says no module in and file exist goto no module needed.
<darkpsi>as i understand it i dont need to load the module since it is now compiled in the kernel but guix system reconfigure is deamanding it to be loaded
<zybell_>And that check must be replaced with one that works with compiled in too. Modern drivers in linux define files in /sys when they are loaded
<zybell_>That works whether they are module or compiled.
<darkpsi>yeah probably going to put a bug report out probably not that many people using guixSD and a nvme
<darkpsi>even then though i think there should be a way to disable the check
<zybell_>so you must test it anyway (no others around) why not first make it work, then present solution with report?
<zybell_>disable will only lead to a next bug report.
<darkpsi>yeah going to do some more research i am new to guix but i have a fair amount of linux and a good programming background
<mbakke>darkpsi: Are you using the stock GuixSD kernel or do you compile your own?
<zybell_>when it fails later on because sth is *not* checked
<darkpsi>standard guixsd kernel
<mbakke>It sounds odd that you would get an error on reconfigure with the stock kernel. Can you paste the full error message to
<zybell_>A post to the ML is linked in log (yesterday I believe)
<zybell_>config_nvme is no longer allowed to 'm' so default is chosen 'y'
<JorgeMorais>Hi. I want to install afew 1.3 via Guix, but Guix currently only has afew 1.2. niebie provided me with but I do not know how to telltell guix to use it. I have already spent hours reading documentation. I have created directory `~/guix_packages/', and stored the package definition in a file named `afew.scm'. Then running `GUIX_PACKAGE_PATH=/home/jorge/guix_packages guix package --dry-run -i
<JorgeMorais>afew' gives error
<mbakke>I found this thread by searching my email, but it's fairly old at this point:
<JorgeMorais>guix package: warning: failed to load '(afew)':
<JorgeMorais>Unbound variable: package
<JorgeMorais>I have also tried naming it `mail.scm'.
<darkpsi>downloading the new kernel seeing if it has been fixed on 4.16.2
<darkpsi>i will report back when it is done
<mbakke>JorgeMorais: You'll also need to override (source ...) with an updated checksum.
<mbakke>You might also need some more module imports. checks own GUIX_PACKAGE_PATH repo
<JorgeMorais>mbakke: I did not understand the second sentence
<mbakke>JorgeMorais: I would recommend changing the git repository directly and submit a patch however, using the "./pre-inst-env" script (see "Running Guix before it is installed" in the manual).
<JorgeMorais>What would be the advantage of that?
<JorgeMorais>I mean, i do not feel confident to submit a patch at this point
<zybell_>in kernel config CONFIG_NVME is no longer allowed to 'm' so default is chosen 'y' explained on guix-devel@
<mbakke>JorgeMorais: Since you get 'unbound variable package', it suggests that you need to import the module that defines 'package'. From a `git grep` it seems you need to add '(use-modules (guix packages))' at the top.
<mbakke>Also, editing the git repository directly may actually be easier than adding your own variant on GUIX_PACKAGE_PATH for a simple package update.
<mbakke>Plus, it would benefit other users.
<JorgeMorais>OK, I will try that.
<darkpsi>mbakke: this is the error
<mbakke>darkpsi: Can you also paste your config.scm?
<darkpsi>a tad messy atm
<darkpsi>the uuid drive partition in question is on a nvme
<darkpsi>and boot part is on another drive and partition
<darkpsi>other then that not that exotic
<mbakke>darkpsi: Excellent, thanks. This is definitely a bug. Which kernel are you running currently?
<darkpsi>4.14.3 and wanting ot update to the latest 4.16.2
<darkpsi>not a problem one 4.14.3
<mbakke>It seems the error message comes from inspecting 'module.alias' on your current kernel, and checking that the system configuration is okay:
<zybell_>How about adding nvme to %redundant-linux-modules?
<darkpsi>ah ok so could i spoof module.alias?
<zybell_>Since 4.16 in kernel config CONFIG_NVME is no longer allowed to 'm' so default is chosen 'y' explained on guix-devel@
<darkpsi>as a quick dirty fix
<mbakke>darkpsi: What error do you get if you add nvme to initrd-modules?
<mbakke>I guess it won't be found, hm.
<zybell_>Not found because not built ^^
<mbakke>darkpsi: As a workaround, you can try "guix system reconfigure --skip-checks ...".
<darkpsi>ah thanks
<mbakke>That should disable the nvme check.
<mbakke>It won't be necessary after you get kernel >= 4.15.
<darkpsi>the error when included nvme is included is the second error in the first post
<darkpsi>thanks i will try --skip-checks
<mbakke>Oh right, I see it now.
<darkpsi>ok bootloader is not installing propely but got past the nvme check so i should be able to sort it out from hear thanks
<mbakke>excellent :)
<soundtoxin>Can anyone here give me a quick rundown on how to mount my Android phone's storage and move files on/off it on GuixSD?
<siraben>There's no openjdk in the guix package list?
<vagrantc>i've been running a system for a while, and have something like 80 generations over time ... how do i remove system generations so they can be garbage collected?
<vagrantc>with guix package --delete-generations= ... i can do that for an individual user's profile
<siraben>guix gc
<siraben>Does that not work?
<vagrantc>but how do i delete the old generations
<siraben>They take up space?
<siraben>I'm new to Guix
<vagrantc>i run "guix gc" and system generaion is still there
<vagrantc>the original installed system generation, that is
<siraben>Guix should have a limit on the generations
<siraben>join c
<siraben>Oops I meant that as a command
<siraben>for IRC
<vagrantcish>i've got apparently 4.9GB of old linux-libre's installed, for example ...
<vagrantcish>what's the equivalent for "guix package --delete-generations=" for "guix system ..." ?
<lfam>vagrantcish: For now, it must be done manually by deleting the gcroots such as /var/guix/profiles/system-N-link
<sneek>lfam, you have 1 message.
<sneek>lfam, efraim says: btrfs and /tmp is refering to some packages like syncthing failing tests on btrfs, but not having issues when CoW is off
<vagrantcish>lfam: so just deleting the undesired symlinks?
<lfam>sneek: later tell efraim: Interesting, I hadn't found that one yet. It should be reported upstream
<lfam>vagrantcish: Yes, taking care not to delete the current one :)
<vagrantcish>lfam: ok.
<vagrantcish>i suppose that's why it's not documented in the manual? there's no way within guix to do it, so ...
<lfam>We should just document the workaround. There was some work-in-progress on this, but it seems to have stalled
<lfam>vagrantcish: Last update, AFAIK:
<vagrantcish>trying to add a libvirtd service... reading
<vagrantcish>i'm not sure where exactly to put (service libvirt-service-type ... in my config ... i've got a (services (cons* (screen-locker-service ...
<vagrantcish>but when i add it there, i get /home/vagrant/config.scm:77:13: libvirt-service-type: unbound variable
<vagrantcish>hint: Did you forget a `use-modules' form?
<vagrantcish>i still really struggle with where things are supposed to go in config.scm
<vagrantcish>i've got (use-modules ... (gnu services virtualization))
<vagrantcish>hrm. apparently i needed to add (use-modules ... (gnu services virtualization) ... as well as (use-service-modules virtualization) as well as (use-package-modules virtualization)
<vagrantcish>or something
<vagrantcish>it's a bit confusing to know when i need use-modules, use-service-modules or use-package-modules
<userr>Are there any Rogue‐like games (except for tome4) in Guix?
<thomassgn>userr: cataclysm dda is a very nice roguelike
<thomassgn>but the settings is survival in modern world after zombie apocalyps
<thomassgn>hyperrogue is interesting, but not so immersive.
<thomassgn>If you want to play cataclysm with tiles (not ascii/utf art) then install cataclysm:tiles (it's in a different output) and the executabel is cataclysm-tiles
<userr>guix package -i cataclysm:tiles?
<userr>There is no such package here:
<thomassgn>yes, :tiles means you want the tiles output of the package cataclysm
<userr>Why is there no nethack if there is a patch?
<userr> s/a/
<thomassgn>No idea. maybe it is being worked on...
<thomassgn>but this looks like you could put in as a package on your computer...
<janneke>i've been "waiting" to see postfix packaged and happened to notice that we already have an exim service!
<userr>I have this command: gcc -c CGE.c -I"/usr/include/SDL" -ansi -Wall -pedantic -ggdb
<userr>Do I need to fix it to compile it on Guix?
<userr>I have sdl and sdl2 packages.
<Sleep_Walker>well, path won't be correct
<userr>How can I fix it?
<userr>I have fixed it using -I"/gnu/store/…-sdl-union-1.2.15/include/SDL"
<userr>How can I fix this command?
<userr>gcc -o CGE CGE.o -lSDL -lSDL_ttf
<pkill9>hmm my ethernet connection wouldn't work after waking up GuixSD from suspend
<roptat>userr: you can add -L/gnu/store/…/lib
<roptat>or I think you can ensure LIBRARY_PATH contains the path to the library (not sure)
<oleo>i tried the guixsd dvd-image, i can boot into it but i don't get no wifi connection
<oleo>the issue is the driver, i think it's lacking
<oleo>wpa_supplicant -Dnl80211,wexp does nothing
<oleo>both fail
<oleo>iwlist does not list anything
<oleo>i only see a cfg80211 module listed
<oleo>but usually i should also see a mac80211 module
<oleo>it's not there
<oleo>so i don't have the phy driver ?
<pkill9>oleo: which wifi do you have? non-free drivers aren't supported
<oleo>ah ok then, that may be it
<oleo>it's a dell laptop
<oleo>probably intel iwlwifi
<oleo>so it's not free ?
<oleo>i suppose....
<jlicht>hey guix
<pkill9>anyone have any idea how to get electron working?
<efraim>can it be built from source? I think we still have a chromium package in the bug tracker
<sneek>Welcome back efraim, you have 1 message.
<sneek>efraim, lfam says: Interesting, I hadn't found that one yet. It should be reported upstream
<efraim>i read through the history, I assume that should be reporting btrfs and /tmp and syncthing
<Sleep_Walker>ACTION has first attempt for docker ready :)
<efraim>It uses go?
<Sleep_Walker>I hate synopsis and description :E
<atw>There's a new way to run Clojure: Doing doesn't seem to require Maven.
<atw>Wait, sorry, that's wrong.
<Apteryx>anyone else tried receiving/pushing files over bluetooth using GuixSD? Seems I need an OBEX server like provided by bluez-tools (bt-obex):
<buenouanq>Would someone be so kind as to post examples of their nginx and certbot service definitions please?
<pkill9>curse nodejs
<JorgeMorais>Hi. It seems that guix environment is not working here as documented. I invoked `guix environment --pure --search-paths guix'. However the env variable `GUIX_ENVIRONMENT' has not been defined, and I verified that PATH and PKG_CONFIG_PATH have not been changed.
<Sleep_Walker>JorgeMorais: are you sure that you are inside the shell with new environment?
<Sleep_Walker>JorgeMorais: IOW does `echo $$` (which prints PID of shell process) differs before invocation and after?
<JorgeMorais>Sleep_Walker: No, it does not differ.
<JorgeMorais>What should I do?
<JorgeMorais>Now I invoked `guix environment' without the `--search-paths' option and it seems to have worked. It seems that the `--search-paths' option causes guix to not enter the new shell. I believe the manual should be clearer
<JorgeMorais>Sleep_Walker: It seems for now that the problem is solved. Now I have to lunch (it is 14:10 here). Thank you for your attention.
<thomassgn>would it be super stupid to make a flag for guix environment and guix build that made it generate a .envrc file for use with direnv? seems like direnv is common here, and I use it for everything myself.
<Sleep_Walker>JorgeMorais: I'm sorry, I was just preparing tea
<Sleep_Walker>JorgeMorais: if you think that guix environment behaves differently than documented, please file a bug
<thomassgn>JorgeMorais: Not sure how clear the manual/docs are. But --search-paths should display/print the environment variables of the new environment. It's for use with tools like direnv (BUT direnv doesn't need it anymore, it has a directive 'use guix' which is like calling guix environment ...)
<thomassgn>I see the info page is not clear about what happens when you run guix environment --search-paths and-then-whatever-extra; Does it cancel creating a subshell with env.? From the described behaviour above it looks like it. Or maybe it just prints current env. vars relative to guix?
<davidl>when Im using tramp to open files on my guixSD server Im getting the error "Wrong method specification for 'ssh'". Any ideas why?
<zero21>yesterday I found that nix package is broken, so i think it should be reported but idk where exactly, should i do it at ?
<alezost>zero21: just send a message to <>
<zero21>i ve been trying to find right way to submit bug but found only , should i paste entire install text or not?
<alezost>zero21: I think the more info the better, but you may just write what seems appropriate to you. If any additional info will be needed, you will be asked for it
<alezost>zero21: in case you forget the bug email, you can always find it by running "guix -h"
<alezost>no problem, if you have questions you may always ask here or at <>
<pkill9>whichdoes anyone divide up their user profile into deparate profiles?
<JorgeMorais>Hi. I want to get afew@1.3 through Guix. Unfortunately, Guix only has version 1.2. So I cloned Guix git repository and edited the afew package definition in mail.scm. I then invoked `./pre-inst-env guix lint afew' and `./pre-inst-env guix build afew'. Therefore, the locally compiled guix must have communicated with the system-wide (and binary-installed) Guix daemon. Is that OK? The system wide Guix was updated very recently
<JorgeMorais>through `guix pull && guix package -u'.
<ngz>I reinstalled my system recently and Gnome is no longer able to find guix installed applications in the application launcher. What am I missing?
<jlicht>JorgeMorais: That should not be a problem, unless your guix daemon is quite old
<JorgeMorais>jlicht: Thank you for the answer
<jlicht>I assume you are running binary Guix on top of a non-GuixSD distro, correct JorgeMorais?
<JorgeMorais>jlicht: yes.
<JorgeMorais>I am trying to package afew@1.3 (Guix only has 1.2). It has been a lot of work. I am a newbie.
<jlicht>JorgeMorais: Good to hear :-).
<jlicht>Once you have verified that it builds reproducibly, and hopefully works, you can send it to the guix-patches list.
<jlicht>If you need some inspiration regarding commit messages, check the commit log for examples of the CHANGELOG-style used in the Guix project.
<JorgeMorais>Thank you for that tip.
<jlicht>yw, and good luck