IRC channel logs


back to list of logs

<j3kyl_>hey, any guix package command issue this: `_IONBF' is deprecated. Use the symbol 'none instead.
<j3kyl_>guix .16
<pkill9_>in the manual, it says that "you can insert an inferior package pretty much anywhere you would insert a regular package". I can't use an inferior-built kernel to the 'kernel' parameter in the operating-system declaration, is this because of a fundamental incompatibility with how the 'kernel' parameter handles the kernel package given to it, or has it just not been modified to transparently handle
<pkill9_>inferior-packages like the other parts of guix have (like manifests and the 'packages' field of the operating-system declaration, like the manual says)?
<pkill9_>(the specific error I get when trying to build that system configuration is: gnu/system.scm:909:35: In procedure kernel->boot-label:
<pkill9_>In procedure package-name: Wrong type argument: #<inferior-package linux-nonfree@4.18.15 3d4b2a0>
***tilpner_ is now known as tilpner
<pkill9>oh looks like it's not incompatible, it's just the part that generates a boot entry name
***catonano_ is now known as catonano
<pkill9>sweet, that is the only part that needs changing
<pkill9>the checked out version of guix with that part modified to use inferior-package-name and inferior-package-version successfully builds the system configuration
<pkill9>so i think it just needs to check if the package given is an inferior-package, i think that's what the guix manifests does
<brendyyn>Anyone running a Tallos II with a free system written any blog posts about it? i'm quite curious
<j3kyl_>hey, icecat 60 are not correct displaying fonts. But its filled with those utf8 square all over. Guix .16/
<j3kyl_>locale are instaled as emacs, vlc are displaying fonts correctly...
<j3kyl_>font-dejavu installed
<fps>j3kyl_: try running fc-cache -f
<fps>after installing some fonts
<brendyyn>so changing the order of the inputs changes the resulting output it seems
<fps>brendyyn: in a package definition? yeah, i suppose
<brendyyn>i ran diffoscope after changin the order and everything contained different store references
<brendyyn>I've found myself making use of sp-convolute. I'm a real lisper now.
<brendyyn>Why are my messages showing Message not available on the mailing list. i must be doing something wrong in thunderbird
***benny- is now known as benny
<vagrantc>i'm trying to add support to be able to make a bootable image for aarch64, but it hangs on starting qemu with qemu eating 100% CPU
<vagrantc>looks like it's passing the wrong console= argument
<vagrantc>ah, think i found it.
<asterope>Can I add custom channels system-wide? I found only per-user solution in the docs...
<vagrantc>hrm. seem to have found a bug in the installer building code ... it just smashes together the console arguments and passes qemu console=ttyS0ttyAMA0
<vagrantc>what boggles my mind is that doesn't change if i set a different console for target-arm32 and target-arm64
<vagrantc>hrm. not as simple as that
<g_bor>hello guix!
*vagrantc waves
***rekado_ is now known as rekado
<rekado>cuirass on berlin is stuck
<rekado>did anyone here get an EOMA68-A20 board from Luke for GuixSD porting work?
<vagrantc>i'll in theory get one soon
<roptat>hi guix!
<rekado>Luke wrote that a “key developer” got a board at FOSDEM.
<rekado>I don’t know who that would be.
<j3kyl_>roptat: o/
<j3kyl_>fps: fc-cache did not correct it... strange
<fps>j3kyl_: i had the same problem when i first installed icecat.. i installed some fonts and then ran fc-cache -f and after that the fonts showed up
<j3kyl_>fps: installed some more fonts font-gnu-freefont-ttf
<j3kyl_>its icecat 60.... i want to use :D
<j3kyl_>fps: it keeps warning me to install locale but IT IS installed
<brendyyn>im always getting a blackscreen when i reconfigure and i have to force a reboot. perhaps we should have a feature to reconfigure such that the new configuration is available in grub on reboot, but not instantiated immediatle
<civodul>Hello Guix!
<rekado>civodul: Hello!
<rekado>civodul: cuirass on berlin is dead, but shepherd isn’t very responsive at this point
<janneke>Hello civodul!
<vagrantc>hrm. trying to build an aarch64 installer image ... basically added target-arm64 almost wherever there was a target-arm32 doing essentially the same thing ... but it just hangs on the qemu step
<civodul>rekado: oh right
<civodul>i'll take a look
<civodul>rekado: as discussed the other day, PID 1 is stuck in read(0...) due to the failed file system mount
<vejetaryenvampir>Triying something new on tmux sorry.
<civodul>heh :-)
<civodul>rekado: i'll try something, but are you able to reboot it if needed?
<civodul>like, if things go wrong
<civodul>vagrantc: it should be enough to run "guix system disk-image -s aarch64-linux install.scm", no?
<civodul>(provided you're already on aarch64 or have offloading set up)
<vagrantc>civodul: do you have to specify -s aarch64-linux ?
*vagrantc will try again
<civodul>rekado: fixed, no need to reboot :-)
<civodul>vagrantc: it's implicit if you're running on aarch64-linux, but otherwise yes
<vagrantc>civodul: will try it again, but i'm pretty sure i wandered down this path because it wasn't working :)
<vagrantc>looks like "guix pull" doesn't yet have any generation management? just manually move around the symlinks?
<civodul>it does have it!
<civodul>well you have to use "guix package --delete-generations=xxx -p ..."
*vagrantc squints
<civodul>IOW, i guess you were right, to some extent :-)
<vagrantc>a little hard to remember, but it makes sense :)
<civodul>yeah we should probably simplify that
<vagrantc>Formatting '/gnu/store/xazy79p18pqs6gq032y0miqfh883kc4z-disk-image', fmt=raw size=1460894966
<vagrantc>qemu-system-aarch64: -nographic: No machine specified, and there is no default
<vagrantc>it has to pass -M virt ... just like armhf
<civodul>oh i see
<civodul>well maybe target-arm32? should be target-arm?
<vagrantc>it's possible that arm32 and arm64 differ
<vagrantc>but currently, yes, i've gotten to the point of qemu stalling while creating the image by just copy-and-paste target-arm32 and making a corresponding target-arm64
<vagrantc>there's definitely something wrong with the console handling:
<vagrantc>- " console="
<vagrantc>- (if target-arm32? "ttyAMA0" "ttyS0"))
<vagrantc>just passes console=ttyAMA0ttyS0 to qemu
<vagrantc>rather than what i think was intended as console=ttyAMA0 console=ttyS0
<vagrantc>e.g. both arguments
<vagrantc>ah, target-arm32 is used for other things as well
<vagrantc>so couldn't just repurpose it
<vagrantc>really loving --switch-generation with guix pull...
<civodul>that could be a newtonmas present :-)
<vagrantc>civodul: thanks for introducing the concept of newtonmas :)
<pkill9>does elogind fail to build for anyone pretty much latest guix? (fails for me on guix commit b9103c827c605dee32baf62816a0429543b3e451)
<pkill9>disregard that, it\s using my custom version of elogind which tries to patch a now nonexistent
<pkill9>hmm I get a hash mismatch from the substitute for nss-certs
<pkill9>downloading from
<pkill9>sha256 hash mismatch for /gnu/store/xbj4fhad0lnz0ziflwi90gyqbls8ains-nss-certs-3.39:
<jlicht>hey guix
<jlicht>late congratulations on the 0.16 release to all those involved still \o/
<fps>pkill9: was the same for me.. someone sneaking in bad certs?
<pkill9>the --fallback flag fixes the issue
<pkill9>well, the issue of not getting the package, not the issue of possibly someon sneaking in bad certs :) it's probably a technical issue though
<pkill9>kudos for all the work on inferiors, now I can upgrade the system without recompiling my modified kernel when any of it's dependencies change :D
<jlicht>pkill9: could you explain how you do this? One of my machines also has a custom kernel, and life is too short to be compiling kernels every week
<pkill9>yes! I'll write a guide, it currently requires modifying guix because the part where it generates the name for the boot entries fails with inferiors, but a simple check to see if the kernel package passed is an 'inferior package' or a normal package, and using the functions for inferiors will fix it
<jlicht>that seems interesting, I'll gladly read it once you have something written down
<janneke>aaarrrrgh, how do i print (with cups) i don't remember
<cbaines>janneke, I fought with cups this morning as well, I did win eventually though :)
<cbaines>how far have you got?
<janneke>there is no "default printer", trying localhost:631 but it does not agree with my user/passwd
<cbaines>I managed to setup a printer through the lpadmin, lpstat and lpinfo commands
<cbaines>I tried the web interface, but it didn't work, it seemed to be trying to write to the store
<cbaines>I also had to use the cupsaccept and cupsenable commands to actually get it to work as well
<joshuaBPMan>Hello, I'm having a hard time getting mcron to work. I'm getting this error:
<joshuaBPMan>mcron: Cannot read files in your ~/.config/cron (or ~/.cron) directory.
<joshuaBPMan>my cron jobs are in ~/.config/cron/
<joshuaBPMan>I can read the files without issue. Why is mcron having trouble reading them?
<janneke>cbaines: how did you find the name of the printer for ldadmin?
*janneke never had problems with the web interface until now
<cbaines>Ah, I may be misremembering the commands, it's actually lpadmin
<cbaines>I think I created a new printer using the -p -m and -v options
<enderby>is there a way to make a certain substitute server the default? for example i want to change to using only, but unless i specify --substitute-urls, it still defaults to hyra
<cbaines>Something like: lpadmin -p test3 -m drv:///hp/hpcups.drv/hp-envy_4520_series.ppd -v hp:/usb/ENVY_4520_series?serial=TH62H3J0VV0660
<cbaines>(where test3 is the name I was giving the printer)
<cbaines>enderby, the build daemon can (and probably is) configured with a list of substitute servers, and I believe it uses them in order, so the first is the default
<cbaines>are you using GuixSD, or another distro?
<enderby>cbaines: guix on debian
<cbaines>enderby, ok, so you may be running the guix-daemon as a systemd service, does that sound right?
<enderby>cbaines: yeah that's right
<ennoausberlin>Hello guix
<cbaines>enderby, ok, out of interest, what does the ExecStart line say in the service file you're using?
<enderby>ExecStart=/var/guix/profiles/per-user/root/guix-profile/bin/guix-daemon --build-users-group=guixbuild
<cbaines>enderby, right, so it's using the guix-daemon from root's profile. I think that updating the guix package in root's profile, then restarting the guix-daemon service might help.
<ennoausberlin>I currently run a low spec laptop while traveling. It has only 32 gb disk space. And texlive is quite big. Is it possible to transfer the store to an sd card to safe internal drive space?
<janneke>cbaines: thanks, it worked!!!
<ennoausberlin>btw guix 0.16 works great
<cbaines>janneke, great :) though I'm really looking forward to when the Gnome printer config stuff works
<enderby>cbaines: ah, yes i think that will do it, thanks. i thot guix pull got the latest guix, i was mistaken
<cbaines>ennoausberlin, are you using Guix as the OS, or on top of another distro?
<ennoausberlin>cbaines: guix sd installed yesterday
<cbaines>enderby, guix pull works per user, so to update the guix package in root's profile, you'll need to run guix pull as root, then guix package -u guix also as root
<cbaines>ennoausberlin, ok, so moving the store to an sd card is probably possible, but it'll require a little care, as guix reads from the store during booting, so if you remove the sd card, the system won't boot
<pkill9>ennoausberlin: out of interest, what laptop are you using for travelling?
<ennoausberlin>sd card is inside permanent. internal slot
<ennoausberlin>asus e200h.
<ennoausberlin>2 gb ram. but X with exwm and next browser works fine
<cbaines>ennoausberlin, ok, I think I've done this before, but I'm no expert. I think I copied the contents of the store to the other device, then changed the mount configuration, reconfigured, checked it worked, and then removed the store data from the original device.
<cbaines>One neat trick I picked up is that once you've got the system booting and mounting the store from the sd card to /gnu/store, you can bind mount /gnu to somewhere else, say /mnt/gnu to allow access to /gnu/store on the internal storage, so you can delete it to free up space.
<ennoausberlin>cbaines: Do I have to start copying from a live boot or stop guix daemon? I am a little nervous because I dont want to break something
<cbaines>ennoausberlin, as the store contents is immutable, it should be safe to copy it while the system is running
<cbaines>I'd avoid running any commands that would change the store contents though
<cbaines>while you're doing the switch
<ennoausberlin>cbaines: I thought so, but was not sure. I am not a long time guix user yet :)
<cbaines>providing you don't delete anything from the internal storage until you've got it working with the sd card, you should be able to boot previous system generations if something goes wrong
<ennoausberlin>cbaines: Ok. I will give it a try. Thank you for your suggestions
<cbaines>ennoausberlin, you're welcome, good luck :)
<ennoausberlin>Is there an easy way to get the internal ath10k wifi to work? At the moment I use an USB wifi extender, but that blocks one usb port and I have to boot ubuntu first because usb_modeswitch is missing on guix I guess
<ennoausberlin>Well, I start copying now. :)
<janneke>cbaines: i have dropped even gnome and use exmw...although i'd be happy if gnome-print solves this for me
***slyfox_ is now known as slyfox
***aweinsto1k is now known as aweinstock
<samplet>janneke: Regarding Gash’s sed, do you mind if I take a crack at adding the missing commands that it needs for gzip?
<rekado>civodul: thanks for fixing it. I’m curious to learn how to fix PID 1 while it’s running / stuck.
<janneke>samplet: that would be grand!
<janneke>samplet: note that gash is not really "my" project, it's rutger's child and his --to-be-- first tangible contribution to free software
<Sleep_Walker>I'm a bit confused with setting environment variables - where should I do it?
<Sleep_Walker>I was considering some ~/.profile which would suite it best - source it once during login
<Sleep_Walker>do you think that /etc/profile could contain in the end something like
<Sleep_Walker>[ -r ~/.profile ] && source ~/.profile
<Sleep_Walker>as I miss it and I haven't find information about how to do that properly - I'm doing it probably wrong :)
<efraim>on guixsd /etc/profile already sources the user's .guix-profile/etc/profile, so i think that would be good
<efraim>it would be nice if it just worked nicely
<efraim>i haven't seen where .bashrc is supposed to source .bash_profile
<efraim>not sure how it would interact with guix environment
<Sleep_Walker>loading this is mess
<Sleep_Walker>interactive shell vs login shell
<Sleep_Walker>my X session is started via non-interactive shell script using #!/bin/sh hashbang
<samplet>janneke: Great! I will report back if I make some progress.
<Sleep_Walker>~/.guix-profile/etc/profile is Guix generated
***Server sets mode: +cnt
<grewil>Newbie question: I wanted to start over with a fresh guix package manager on my Debian (as I accidentally installed a huge number of packages). I deleted /gnu and /var/guix, then did a new binary installation according to instructions. But when I did "guix pull" in the new installation, it started to install the huge pile of packages from the previous installation. What do I need to delete apart from /gnu and /var/guix in order to
<grewil>restart from scratch?
<mbakke>grewil: guix pull needs to download a large number of packages on the first invokation, are you sure that wasn't it?
<g_bor>hello guix!
<mbakke>I don't think it's possible for Guix to know about your previous profile if you removed /var/guix and /gnu.
<mbakke>Hi g_bor :-)
<grewil>hm, but it is still compiling after halv an hour, it has compiled gcc, ed etc.
<mbakke>grewil: Did you remember to authorize the substitute server?
<grewil>as far as I remember, ed was something that was not installed at the initial installation, but a package that I chose to add later. Now it is reappearing in the new installation.
<grewil>hm, no I did not authorize the server again, you're right - I'll start over and this time I'll authorize.
<g_bor>I've run into something.
<g_bor>I tried to get a working pinentry for my ssh logins.
<g_bor>I've been using pinentry-gtk-2 until now, but it did not work in this situation.
<g_bor>I had to modify my pinentry-program, but I also had to set GPG_TTY to get it working.
<g_bor>this is actually stated in the gpg doc, but it is easy to miss that as pinenty-gtk-2 works just fine.
<g_bor>Anyone has more insight?
<g_bor>The gpg doc also says that this value should be set inside gui also.
<efraim>GPG_TTY=tty or something like that?
<g_bor>efraim: yes, actually GPG_TTY=$(tty) worked fine
<efraim>I have that in my .profile on my headless machine
<mbakke>grewil: You don't need to start over, you can authorize the server at any time :-)
<grewil>aha! thanks!
<marusich>g_bor, FWIW, I use 'pinentry-program /home/marusich/.guix-profile/bin/pinentry-gnome3' in my ~/.gnupg/gpg-agent.conf
<marusich>It seems to work everywhere.
<marusich>However, there is a significant delay (many seconds) when running it from a virtual terminal. It seems to fall back to some sort of text-based interface, but it still works in the end.
<marusich>I've seen the GnuPG documentation that mentions GPG_TTY, but honestly I've never understood what it was trying to tell me. I've never had to set it.
<efraim>marusich: what does ls -l $(readlink /home/marusich/.guix-profile/bin/pinentry-gnome3) say?
<marusich>it says: /gnu/store/2dxp1j1hwfj9801l56zzkj7m358mq4rs-pinentry-gnome3-1.1.0/bin/pinentry-gnome3
<marusich>which itself is an ELF executable
<efraim>so maybe it fails to open pinentry-gnome3 in the terminal and falls back to pinentry-curses or pinentry-tty
<efraim>on one headless box its pinentry->pinentry-efl and the other box its pinentry-tty, and the tty one takes no time
<grewil>mbakke Interesting, it seems I cannot successfully authorize the default substitute when I use the binary installation file for 0.16.0. If I use 0.15.0, it uses the substitute server.
<g_bor>ok, thanks.
<grewil>mbakke Or I'm doing something wrong, which is likely.
<g_bor>Now my pinentry program is simply pinentry, and it gives a text based interface, nut I have to set GPG_TTY to get it working.
<g_bor>I guess I can try the gnome3 one also later.
<rekado>oh, is (a part of) savannah down?
<rekado>I get a 502 for this URL:
<g_bor>rekado: same here
<civodul>rekado: re PID 1: "sudo gdb -ex 'attach 1'", and then fearlessly type "return 0" on the right frame :-)
<civodul>"don't try this at home" as they say
<rekado>so… with savannah down plain “guix pull” won’t work.
<civodul>you could provide a different URL, including file://
<civodul>but yeah, that makes things more difficult
<g_bor>rekado: I did not try pulling, just checked the url.
<civodul>rekado: and FWIW the above URL is 200 for me
<rekado>it’s still 502 for me :-/
<g_bor>rekado: now it seems to be also ok for me
<mbakke>Who maintains ?
<g_bor>but respose time was very bad, I thought it will time out
<mbakke>grewil: Are you still not getting substitutes?
<rekado>mbakke: don’t know. Lots of mirrors and all not updated since 6 months.
<g_bor>mbakke: gitlab says this about members:
<grewil>mbakke I downgraded to 0.15.0 where it works (for me - I may be doing something wrong with 0.16.0).
<mbakke>grewil: What went wrong with 0.16?
<grewil>the authorization step seems to fail, because Guix won't use the substitute servers even after running the guix archive --authorize command.
<grewil>Perhaps I should try once more, just to figure out if there can be another explanation.
<rekado>g_bor: now I just get “guix pull: error: Git error: early EOF”
<rekado>something’s still fishy
<g_bor>unfortunately I don't have a recent checkout to publish somewhere :(
<efraim>I have at
<rekado>g_bor: that’s no problem
<rekado>I’m just testing a change to “guix pull” and noticed this.
<g_bor>I also have guix, but not the maintenance repo...
<rekado>for that I just followed a link in the rebuilt website and I wondered if this was somehow my fault :)
<g_bor>ok, I see now
*rekado fixes supercollider
<mbakke>Any idea why Berlin failed to evaluate b52bcb2?
<g_bor>mbakke: no idea here, but this looks very different from the rest, so the display code could hide some clues here...
<g_bor><span class="oi oi-x text-danger" title="Failed" aria-hidden="true"></span>
<g_bor>this is displayed there
<g_bor>is the cuirass repo available, or is it also failing?
<rekado>g_bor: it’s here:
*rekado –> zzZZ
<g_bor>mbakke: yes, cuirass repo is available
<g_bor>this icon is rendered when the evaluation produced zero build jobs.
<g_bor>relevant code is:
<g_bor>mbakke: the issue is tracked here
<mbakke>g_bor: Thanks!
<civodul>so, for tomorrow evening, someone proposed which has veggie options and looks good
<g_bor>civodul: have a nice time!
<g_bor>I can only make it on Monday afternoon, will try to get there as soon as I can
<vagrantc>civodul: looks promising ... which one?
<grewil>mbakke just in case you're interested. I have now tried again with both 0.15.0 and 0.16.0, and I can only get the substitutes authorized in 0.15.0. My reinstall script is here: - afaik it should be correct enough to work.
<civodul>vagrantc: the one rue d'Alibert looks like the closest one
<civodul>they have veggie options, not sure if it's gluten-free though
<mbakke>grewil: For 0.16, the key that needs to be authorized is "~root/.config/guix/current/share/guix/"
<mbakke>Throwing Glib into the mix only adds 200 rebuilds for current staging.
<grewil>mbakke aha, thanks! I should try update the manual, it still says the old one.
<mbakke>grewil: Hope it works! Note that once you've run "guix pull", it doesn't matter if you started from 0.15 or 0.16.
<grewil>mbakke cool! I had not thought of that!
<vagrantc>civodul: it looks good enough for me
<civodul>let's do that
*civodul sends email