IRC channel logs


back to list of logs

<roptat>in any case, the fact that building from the VCS means building without pre-generated files (like configure and such) is a good point for me, so I'd be in favor of switching to VCS whenever possible
<anadon>But that wouldn't allow the same degree of run-time augmentation because it closes off more information from the invokation.
<roptat>I've been rebuilding java packages over and over again all day long, but I'm making progress on my wip-maven-build-system branch
<roptat>it's getting really hot even with the windows open here ^^'
<roptat>oh, found a duplicate package: java-xz is the same as java-tukaani-xz
<KE0VVT>Installing Guix on "scruffy".
<jonsger>KE0VVT: what is scruffy
<KE0VVT>jonsger: HP desktop. 7 GB RAM. 160 GB HDD. Core 2 Duo. 19 in LCD.
<mbakke>nckx: thanks for taking care of zziplib, I thought that was the big remaining python2 dependency, but according to 'guix refresh' there are still 7540 packages that depend on python2 one way or another
<KE0VVT>System is taking a LONG time at "Loading /gnu/store/...-system/boot".
<mbakke>we should reduce that number to close to 'guix package -A python2 | wc -l' for the next core-updates round
<ptibedo>Re ! I create a recipe for the package gnome-builder. How I must use it ?
<mbakke>ptibedo: how did you create the package? if you have it in a file, make sure the last line of the file is 'gnome-builder' (i.e. the variable name) and run 'guix build -f the-file.scm'
<ptibedo>not something we can build.
<ptibedo>can you help me ?
<ptibedo>can I send me recipe ?
<pkill9>we need guix on our phones
<pkill9>we need it with us at all times
<roptat>ptibedo, can you share your file through
<roptat>you need to give as a link :)
***KE0VVT49 is now known as calher
<roptat>as mbakke said, put gnome-builder in its own line at the end of the file
<calher>Man, it takes a long time to "guix install polari".
<calher>Using Kiwi IRC in Web right now.
<roptat>define-public defines a variable, and doesn't return anything (it returns #<unspecified>), and guix cannot build that. define-public creates a variable called "gnome-builder" which contains the package definition. putting the name of the variable at the end gets it evaluated and a package object is returned
<calher>Successfully running encrypted GNOME desktop with Guix System on scruffy.local.
<calher>Is it possible to set up a DLNA server?
<ptibedo>what . the define-public I must put end file ?
<roptat>ptibedo, non juste gnome-builder tout court
<ptibedo>niquel merci !
<roptat>define-public défini une variable, ça renvoie rien de spécial, juste #<unspecified>, donc en rajoutant gnome-builder à la fin, tu renvoie le contenu de la variable que tu viens de définir
<ptibedo>but been failed XD
<roptat>une autre solution, c'est de virer le define-public englobant
<ptibedo>je débute donc je mettrais constament à la fin la variable à la fin
<roptat>(mettre le nom du paquet à la fin c'est un peu comme x = 1; return x dans un langage impératif, je trouve pas ça très joli :p)
<ptibedo>il me dit qu'il ne reconnait pas build-plain
***cal is now known as calher
<ptibedo>je dois construire avec meson et ninja
<ptibedo>dans le champ arguments
<greyseek3r>Hi all. I installed Guix System in a VM. After a global update with "sudo -i guix pull && sudo -i guix system reconfgure /etc/config.scm" xorg is not rendering fonts att all. Any ideas?
<calher>Does it go against the FSDG, to advertise Facebook at the startup of GNOME?
<nckx>It's just dodgy.
<mbakke>greyseek3r: you need to install some fonts, there is a regression in recent Guix
<roptat>ptibedo, qui te dit ça ?
<roptat>guix ou la phase configure de compilation ?
<ptibedo>la phase configure
<calher>What program is recommended for serving media to the home by DLNA protocol?
<ptibedo> ERROR: Pkg-config binary for machine MachineChoice.HOST not found. Giving up.
<calher>I found few options with "guix search DLNA".
<ptibedo>trouve pas meson apparamment
<nckx>greyseek3r: Try installing font-gnu-freefont, it should at least give you text to read.
<roptat>essaye d'ajouter pkg-config dans native-inputs
<greyseek3r>Should I install it globally with "sudo -i" or will it do with a normal user profile?
<nckx>Also, you almost certainly should not be using ‘sudo -i’ but ‘sudo’ unless you are a wizard. It does not sound like you are.
<ptibedo>mais le meson n'est pas installé
<nckx>greyseek3r: And you shouldn't pull as root either.
<ptibedo>comment je dois lui dire ?
<nckx>guix pull && sudo guix system reconfigure…
<nckx>greyseek3r: Installing the font can be done as regular user.
<calher>Hm. It seems Kodi can act as a DLNA server.
<ptibedo>comment je dois lui dire dans la recette pour qu'il installe meson
<greyseek3r>Oh. That explains it. I read in the docs that if you want to update the whole underlying thing you can use "sudo -i". And there was not warning about it, so I just did it.
<greyseek3r>Will install the fonts and report back
<calher>But does Kodi's UPnP support conflict with Shepherd?
<roptat>ptibedo, si tu as utilisé meson-build-system, meson est installé dans l'environnement de construction
<roptat>l'erreur a plus l'air de venir du fait qu'en se lançant meson n'arrive pas à exécuter ce qui se trouve dans (il arrive à la ligne 250 quand même)
<nckx>greyseek3r: Read more carefully! The docs don't say that. 😛
<ptibedo>dont pkg-config dans inputs ?
<ptibedo>ou un autre champ qui s'apelle native-inputs ?
<greyseek3r>nckx: Isn't it mentioned in ?
<roptat>un autre qui s'appelle native-inputs
<greyseek3r>nckx: or have I misunderstood?
<roptat>tu vas aussi devoir importer le module (gnu packages pkg-config)
<nckx>greyseek3r: I think that's been through about 3 rewrites or at least proposals and I guess it's still not clear.
<KE0VVT>Mi ne povas paroli bone. \ Je ne peux parler bien.
<nckx>greyseek3r: The first 2 paragraphs (‘Success’ and the two example commands) are quite clear, but then the rest tries to give some background info but appears to mainly confuse new users like yourself.
<nckx>Not good.
***andrej is now known as Guest2065
<ptibedo>et je dois mettre quoi dans le native-inputs ?
<roptat>(native-inputs `(("pkg-config" ,pkg-config)))
<nckx>greyseek3r: I suggest you ignore everything after ‘Security Updates’ TBH.
<nckx>On a ‘regular’ (single user) system that's all you need to commit to muscle memory to update your system.
<ptibedo>il manque libdale
<greyseek3r>nckx: I thought so too. Seems a warning is needed there. Out of curiousity, is it not necessary at all to "sudo -i", even after having the system for say a year or so?
<nckx>greyseek3r: Correct. I've been using Guix System for 5 years. My root user has no Guix profile (→ has never pulled once).
<ptibedo>pour chaque dépendance je dois mettre un gnu package et un input ?
<greyseek3r>nckx: Righto!
<greyseek3r>nckx: I installed, reconfigured and rebooted but still no fonts. Could it be a locale issue? I have both glibc-locales and glibc-utf8-locales installed as well.
<nckx>greyseek3r: I sympathise, and I can (with some effort) see how you might've read that as endorsing ‘sudo -i’. The problem is that it starts quite clear, then someone thought ‘well we have to at least mention root's guix’ and it's not clear to new users that it's just background info for (more) advanced/control-freaky users.
<roptat>ptibedo, au moins un input, éventuellement un gnu package s'il en manque un
<roptat>chaque module peut avoir plusieurs paquets, qui n'ont pas toujours le même nom que le paquet
<nckx>greyseek3r: Hm, shouldn't be locale-related *touches wood*. What do you mean by ‘no fonts’? Absolutely none? Squares? Some letters but not all? A screenshot might be helpful.
<roptat>tu peux utiliser "guix show <le paquet>" pour voir dans quel fichier il se trouve, c'est le gnu package à importer
<greyseek3r>nckx: just squares with numbers in them
<nckx>greyseek3r: Actually, try running ‘fc-cache -rv’ first and restarting your desktop.
<ptibedo>donc je peux mettre plusieur arguments dans un input si j'ai compris ?
<nckx>greyseek3r: & make sure /var/cache/fontconfig doesn't exist but it shouldn't on modern Guix System.
***catonano_ is now known as catonano
<greyseek3r>nckx: it exists, I'll remove it then...
<nckx>Eh, weird, I thought someone had taken care of that while I was out.
<nckx>Thanks for noting that.
*nckx almost off to bed; 2:30 here. If nobody here can help you, greyseek3r, please fire off a bug report to if you can't find your exact issue at
<roptat>ptibedo, oui par exemple (inputs `(("paquet-a" ,paquet-a) ("paquet-b" ,paquet-b)))
<roptat>(avec l'indentation qui va bien)
<greyseek3r>nckx: Will do. Thanks for answering. Here's a screenshot btw
<KE0VVT>scruffy.local crashed after I SSH'd into it.
***jonsger1 is now known as jonsger
<nckx>greyseek3r: That's after deleting /var/cache/… & running fc-cache -rv? Hm.
<alextee[m]>i get that font issue with `guix pack` too, when running the packed thing on other distros
<nckx>greyseek3r: Does ‘fc-list’ return any fonts?
<nckx>You can install ‘wgetpaste’ on a VT in the VM and pipe its output through a pastebin, although you might have to use ‘DISPLAY=:0 fc-list’ (never tried).
<nckx>The output from ‘fc-cache -rv’ would also be handy.
<greyseek3r>nckx: I actually couldn't remove /var/cache/fontconfig directory, even with sudo. It warns that it's a readonly file system
<greyseek3r>but fc-list generates a lot of output with fonts and their directories
<nckx>greyseek3r: And just to sanity-check: ~/.guix-profile/share/fonts/ttf/FreeSans.ttf exists, right?
<greyseek3r>Yes, though not in ttf, but in truetype directory. I guess that's what you meant
<nckx>greyseek3r: Yeah, it was renamed this week (by me, so I should've remembered 😛)
<alextee[m]>nckx: is there something i should try too in my guix pack tarball? im trying to run ./gnu/stsore/....fontconfig-.../bin/fc-cache but i get no such file or directory
<alextee[m]>even though the file is there o.o
<nckx>greyseek3r: That read-only /var/cache/fontconfig is puzzling to me. You said you ‘installed Guix System in a VM’. So it's a regular (virtual) disk, right? Not some layered overlay business?
<greyseek3r>Yes. Regular qemu qcow2. Let me run the command again.
<nckx>Whenever ‘~$ ./obviously/there/binary’ returns ‘No such file or directory’ it's time to run ‘ldd ./obviousl/there/binary’.
<greyseek3r>nckx: Ah. Now it says "cannot remove 'fontconfig': Device or resource busy"
<nckx>This all sounds very strange to me but I never use VMs.
<calher>Yeah, the computer crashed once I logged in over SSH and started to tab-complete "gui".
<nckx>Sheesh. Everybody's having a bad night in #guix. :-/
<greyseek3r>The fontconfig directory has the sticky bit turned on too.
<calher>Had to go back downstairs to restart the thing and type in the passphrase twice for the disk.
<calher>I just hope I can get a media server and DVD ripping setup going on here.
<calher>It's what I intalled Guix System for.
*nckx needs to sleep.
<calher>Also, I notice the computer first presents me with a TTY login prompt, and takes a LONG time to get to GDM. If I didn't know any better, I would have logged in at the TTY and thought GNOME was broken.
<nckx>Guix = slow, Gnome = slow, Guix + Gnome = ♥.
<calher>...and slow. 😂️
<calher>Whoa, why are the emoji black and white in Polari? They are not this way in Trisquel.
<nckx>Good night everyone. greyseek3r: I was expecting this to be an easy thing to fix, sorry about that. Do file a bug if you get nowhere here. Some things don't sound right (well, apart from the obvious facts that fonts should not be square, I mean).
<calher>Maybe I should have just stuck with Trisquel 8 for my media server and desktop...
<nckx>calher: guix install font-google-noto if colourmojoes are *that* important to you.
<calher>They are. I'm a picky guy.
<greyseek3r>nckx: I appreciate the help anyway. Will try to see if I can fix this, and probably a bug report with steps on how to reproduce. Good night.
<calher>Hm. Weather app not opening.
<calher>Man. It's so beautiful, but it's so slow.
<calher>I'm not getting any indication when I insert a DVD or CD.
<calher>And Kodi still thinks it's midnight.
<calher>What's the use... I can't do anything on this OS. I can't mount a CD. I can't SSH into it without it crashing.
<calher>Do I need special privileges to use the CD drive or something?
<greyseek3r>calher: You probably need to be a member of cdrom group
<calher>I installed this with the GNOME desktop option. The auto-config.
<greyseek3r>calher: modify your /etc/config.scm and add "cdrom" under supplementary-groups
<greyseek3r>calher: then run "sudo guix system reconfigure /etc/config.scm" then a reboot
<calher>How is this system supposed to be useful if people have to reboot it all the time?
<greyseek3r>Well, a log out and log in might suffice too. Not sure though.
<calher>I've been waiting an hour to install a font. Can't reboot right now.
<calher>I would be calmer if I could SSH into this and sit on the couch while waiting on this, but that crashes the system.
<calher>So I have to hunker down in the office chair.
<calher>Oh, maybe it was just GNOME's Automatic Suspend that interrupted my connection. Oops.
<ryanprior>calher: you don't need to reboot to change your own packages, but I think you do to change the system packages. I never use Guix on Debian, so I never reboot after running a guix command, and rarely after running a dpkg one.
<calher>ryanprior: I'm running Guix System.
<ryanprior>I changed my mind halfway through that sentence
<ryanprior>I started writing "I never tried Guix System" and then went to "I'm using Guix on Debian" and ended up not making sense
<calher>I'm having a bad day.
<nckx>If the laptop can't sleep the nckx can't sleep, and the laptop is still compiling Linux. So here I am for a few more minutes.
<ryanprior>Fair enough =X I see you're a Trisquel user also?
<nckx>calher: You don't need to reboot to update anything but the kernel.
<KE0VVT>ryanprior: I use Trisquel for all my daily stuff. I got a second computer and wanted to try Guix, since I'm just going to use it as a media server.
<KE0VVT>ryanprior: But maybe Guix is too much work.
<ryanprior>I don't think Guix is nearly as friendly as Trisquel yet. It makes you appreciate the huge amount of effort the Trisquel/Ubuntu community put into that software to make it so nice!
<KE0VVT>ryanprior: I mean, who tells the installer to make a desktop configuration and forgets the CD drive?
<ryanprior>I don't have a CD drive on any machine I own
<KE0VVT>Not having a CD drive is fine if all you do is use Netflix, but I don't do DRM.
<ryanprior>I occasionally buy mixtapes and demos on CD from local artists because I want to support them but I havne't listened to a one in the past ~7 years bc no hardware and not enough desire to buy an external drive and have it sitting around =X
<KE0VVT>Sure, you can torrent, but why wait hours when you can rip a DVD from the library for free?
<KE0VVT>Physical media is way faster than the network.
<KE0VVT>Streaming is bad.
<ryanprior>Yeah no argument from me
<KE0VVT>It's just a coverup for DRM.
<nckx>KE0VVT: It's easy to sit around and complain but it doesn't improve anything. If there's a bug with the installer and optical drives (which hardly anyone still uses, ‘DRM’ and ‘Netflix’ are non-sequiturs), report it, or better yet help fix it.
<KE0VVT>nckx: How do people get their entertainment, then? Even free software users seem to be using Roku with Amazon Prime.
<nckx>I don't know how people get their entertainment. I've never used DRM and it cost me 0 effort so it can't be that hard to avoid it.
<nckx>But this is #guix, the ‘everybody here is probably “pragmatic” and streaming Netflix’ argument just doesn't hold up.
<nckx>We're not #opensource.
<KE0VVT>Well, I'd like to know what people's TV setups are. Mine is currently ripping DVDs on my Trisquel laptop with vobcopy and serving them to the Roku with DLNA.
<KE0VVT>Or torrents. Depending on where I get the movie from.
<nckx>I have a 4 TiB home server full of films. We'll not discuss how they got there, but it wasn't using an optical drive. I just use sshfs to watch them on my laptop with MPV, so probably not the kind of ‘set up’ you mean (don't have a TV).
<nckx>What does DLNA give you?
<nckx>The only place I've ever seen that was in my ISP's router settings and that made it sound like a Windows media thing.
<KE0VVT>nckx: DLNA lets all the TV's in the house access the sum of all our DVD collections.
<KE0VVT>DLNA is a protocol. minidlna and Kodi use it.
<nckx>So it's not proprietary garbage, good to know.
<KE0VVT>Whatever you have - Roku, Kodi, some DVD players - can access the server's collection and stream from it.
<KE0VVT>So right now, I have the back-end replaced. Later, I can get a computer for the front end.
<nckx>So it's a network file system?
<KE0VVT>nckx: I guess, but it's made to be understood by dumb media players.
<greyseek3r>Is there a way to persistently define GUIX_LOCPATH? The /etc/profile is read-only. So is the ~/.guix-profile/etc/profile
<KE0VVT>Wow, Gajim is still downloading.
<KE0VVT>nckx: If you don't need to reboot, how do you update a user's groups?
<apteryx>greyseek3r: ~/.bash_profile ?
<apteryx>I'm not sure what you are trying to achieve
<greyseek3r>Managed to reproduce the font issue on a fresh install even without using "sudo -i". Normal guix pull and reconfigure will mess it up.
<nckx>KE0VVT: OK, you've found the single exception, but even that 1) doesn't really require rebooting, just ‘logging in & out’, however that's defined for your system/desktop and 2) has nothing whatsoever to do with Guix.
<greyseek3r>apteryx: I'm trying to instruct the system on where to look for locales, and I want the variable to be read by GDM
<KE0VVT>nckx: I'm using the Guix System. On Trisquel, you just log in and log back out. No rebooting. But how does the system know you changed config.scm?
<nckx>Guix is no different.
<nckx>The system ‘knows’ because you run ‘guix system reconfigure’.
<nckx>It's not a configuration file that's read at boot. You can change your system .scm and keep running with the old settings for a year until you reconfigure.
<KE0VVT>Will running that command wipe out my GNOME desktop, just to rebuild it again, or will it just leave everything in place if none of that part of the declaration has changed?
<nckx>The latter.
<nckx>I run it once a day or more, it just means ‘apply this now’, it's not a reinstallation.
<nckx>I am super tired and my kernel is finally done, good night again and this time for realz.
<KE0VVT>I thought the functional nature of the system meant that everything was redone from scratch all the time, based on what was declared.
<KE0VVT>nckx: Good night. Thank you.
<nckx>KE0VVT: And you good luck 🙂 Functional means same input → same output, so if your Gnome configuration fields don't change your Gnome installation won't change. Guix could wipe all of Gnome from the store and replace it with an identical copy but that's just inefficient, not more functional.
<apteryx>greyseek3r: are you on a Guix System or other distro?
<apteryx>I'm not sure, but I think GDM should source your ~/.profile; you could try adding your 'export VAR=VALUE' there
<greyseek3r>apteryx: I'm on a Guix System. There isn't any .profile, will try to create on and see if it helps
<reepca>hmm, how do I make a guile process single-threaded? We use without-automatic-finalization to temporarily shut off the finalization thread while we call clone in (guix build syscalls), but I still count 5 entries in /proc/<pid>/task
<greyseek3r>I wonder if the whole font issue is somehow a vm/qemu issue? Now I'm installing guix in Virtualbox to see if the fonts will still be messed up after a pull and reconfigure
<calher>Scared to "guix system reconfigure" while "guix install".
<pkill9>is it possible to change the default tty that guix system boots into
<pkill9>for console login that is
<pkill9>i would like it to switch to tty2 so i can avoid all the kernal output clogging up the output
<apteryx>pkill9: there must be a getty service that you can configure
<pkill9>yea there's a default one
<pkill9>i thought there is six of them
<pkill9>but i'm not sure
<pkill9>maybe they get created when you switch to them
<pkill9>to the tty's
<apteryx>pkill9: have a look at the bottom of (gnu services base)
<apteryx>it's defined in the %base-services list
<chipb>pkill9: what about adding 'quiet' to the kernel command line?
<pkill9>hmm, I switched to tty2 and it continued outputting kernel messages there
<pkill9>yea i'll do that
<apteryx>pkill9: which tty is it booting to by default?
<apteryx>greyseek3r: if you're using Guix System, you shouldn't have to mess manually with the GUIX_LOCPATH variable (it should be taken care of automatically for you).
<apteryx>calher: are things slowly progressing?
<calher>apteryx: Yes.
<pkill9>apteryx: tty1
<pkill9>but i dont think changing the tty will fix it
<pkill9>actually it might now i think about it
<pkill9>yes it will, the 'quiet' kernel option works for kernel output
<pkill9>and shepherd writes output to tty1 but not tty2
<apteryx>pkill9: this may be the doing of %default-syslog.conf
<pkill9>oh, interesting
<apteryx>it logs *.alert;auth.notice;authpriv.none /dev/console, which I presume is /dev/tty1
<apteryx>so if this annoys you, you could probably change that line
<pkill9>yea i was thinking that
<pkill9>yea i just wrote to /dev/console and it appeared in tty1
<pkill9>actually no, it appeared in this tty, tty2
<pkill9>i'll try changing it anyways
<greyseek3r>apteryx: I'm having the issue with font rendering. I see an error message when I do a reconfigure for the first time. Guile says something about being unable to install locale. I don't know if this is what's causing the font issue
<calher>Hm. Cannot run Telegram client.
<calher>"No such file or directory."
<apteryx>greyseek3r: if your user profile is newer than your system profile, or vice-versa, it may cause issues with locale. On my machine, echo $GUIX_LOCPATH -> /run/current-system/locale. So it's bound to the system profile.
<apteryx>If you're having locale issues, try having both in sync or close enough
<apteryx>(your system & your user profile)
<apteryx>calher: which guix package is this?
<calher>This is upstream tarball binary for GPLv3 client Telegram Desktop.
<apteryx>hmm, running binaries linked to FHS located lbraries will not work on Guix (given that every libraries are under /gnu/store..., e.g., not following the FHS)
<apteryx>calher: if you can get the source of the client and compile it yourself on Guix, that should work.
<calher>OK. It uses Qt.
<apteryx>Then better even, crystalize it as a Guix package.
<apteryx>calher: there's qt in Guix. Qt5 is called qtbase IIRC.
<calher>Speaking of Qt, I know Guix has Mumble. I should install it.
<apteryx>Recently Linphone was also added for telephone needs
<apteryx>It works well.
<calher>Weird... I'm getting lots of downloads just from "guix system reconfigure", when all I want to do is change my user group.
<apteryx>calher: have you guix pull'd since the last time you ran it?
<calher>No. :(
<calher>Can I ^C this?
<apteryx>you can always interrupt Guix processes safely.
<apteryx>the transactions are atomic, so they either modified your system state or didn't.
<apteryx>greyseek3r: also, you could try installing fonts like 'font-dejavu'; I think this worked around fonts issues for others in the past
<calher>What is needed to play MP4? openh264?
<greyseek3r>font-dejavu actually solved it. Damn.
<apteryx>calher: mpv should be able to do it.
<greyseek3r>Perhaps it should be installed from the beginning.
<apteryx>greyseek3r: lfam was struggling with the same issue recently
<apteryx>lfam: ^^
<apteryx>greyseek3r: it's a bug. There's supposed to be a fallback kicking in, but it doesn't in some cases, for some reason.
<apteryx>greyseek3r: this is the one
<calher>apteryx: OK. I was trying to open it in Videos.
<apteryx>calher: I'm not very familiar with the GNOME environment (I use a rather barebone setup)
<apteryx>hopefully a GNOME user will tip in
<greyseek3r>I'll comment on the issue and mention the workaround
<greyseek3r>apteryx: out of curiousity, what does "localectl" show on your system?
<greyseek3r>apteryx: If it's a Guix System that is
<apteryx>I don't have that command installed
<apteryx>which package is it from?
<apteryx>I'm using the Guix System, yes.
<greyseek3r>Weird. Usually the *ctl commands are systemd components, and I was surprised to find it in Guix.
<greyseek3r>"whereis localectl" gives me this: "localectl: /gnu/store/[base32hash]-profile/bin/localectl
<apteryx>try: realpath `which localectl`
<apteryx>ah, it must be from localed
<apteryx>which is a tiny systemd daemon used by GNOME or something
<apteryx>so, I'm not running that daemon on my system. I don't think that command would print something valuable on my machine,but I can try if you think that can be useful.
<apteryx>well, if I run it from the store, it says: Could not get properties: The name org.freedesktop.locale1 was not provided by any .service files
<apteryx>yeah, it's not installed so its DBUS interface hasn't been registered with my dbus service.
<KE0VVT>apteryx: For evangelism purposes, I use GNOME. I used to use i3.
<KE0VVT>Media server on scruffy.local is a success!
<KE0VVT>It even comes with the added improvement of metadata!
<KE0VVT>Now, onto ripping my DVD collection.
<apteryx>KE0VVT: neat :-)
<KE0VVT>In summary:
<KE0VVT>Bradley Kuhn: Confessions of a command line geek: why I don't use GNOME but everyone else should
<KE0VVT>By the way, the default GNOME desktop configuration given by the installer does not allow the web browser to play MP4.
<ryanprior>Is there a way to install Python such that python is python3
<ryanprior>in other words `guix environment ?? -- python --version` -> Python 3.8.2
<apteryx>ryanprior: python-wrapper
<ryanprior>Perfect thank you apteryx
<rekado_>calher: Gnome Videos probably uses Gstreamer, so you’ll need to pick gstreamer plugins.
<KE0VVT>rekado_: Thanks. "guix search" doesn't tell me if it's installed.
<rekado_>“guix search” searches packages. It does not show you installed packages. You can use “guix package --list-installed” for that.
<ryanprior>I'm getting a warning saying "consier running guix pull and guix package -u to get security updates"
<ryanprior>When I run guix pull, it says nothing to be done. When I run guix package -u, it just prints that warning again.
<wxie>KE0VVT: You can install emacs-guix package, and use M-x guix in GNU Emacs to view packages.
<idoesnot[m]>Im trying to write a package definition for a package which only installs auxillary data for another package. The primary package needs to have environment variables defined to tell it where to find the auxillary files which I'm assuming can be done using the (search-paths) part of the package definition. However, when I write a package definition like this and install it with "guix package --install-from-file=foo.scm", the
<idoesnot[m]>environment variable I specified does not show up when using "guix package --search-paths". Is this something particular to the way I installed the package or is it something I've done wrong in the package definition?
<sammich>howdy, is clangd packaged in guix?
<srandom>If it is used for lsp, you can use ccls.
<srandom>name: ccls
<srandom>version: 0.20190823.5
<sammich>Cool, thanks. it seems to be kinda working, do i need anything aside from clang to compile things with it? Its giving me errors for a hello world program
<srandom>If you compile C program in guix, usually use gcc-toolchain,clang-toolchain
<sammich>ah sorry, i was trying to use clang++ but i hadnt installed ld-wrapper
<sammich>its working now
<srandom>In the update, I saw that guix pack -RR added fakechroot, but something went wrong when using it.
<srandom>this is strace's log:
<rekado_>srandom: could you please send this to Please also include the output of “guix describe”
<rekado_>ryanprior: does it also say that your version of Guix is x days old?
<srandom>I will try again if I can solve it, if not, I will send an email
<rekado_>does /home/chromebook/tmp/gnu/store// actually exist?
<olivuser>hello guix!
<olivuser>I am trying to set a xorg-configuration for the slim login manager. Has anyone managed to set another keyboard-layout?
<olivuser>because when I try to use the (keyboard-layout keyboard-layout) form - which works for the bootloader -, it complains.
<olivuser>"Wrong type to apply: #<<keyboard-layout> name: "de" variant: #f model: #f options: ()>"
<srandom>rekado_, It exists.
***janneke_ is now known as janneke
<pinoaffe>hi guix!
<pinoaffe>`sudo guix system reconfigure <path-to-config>` fails with "Throw to key `encoding-error' with args `("scm_to_stringn" "cannot convert wide string to output locale" 84 #f #f)'." - what's going wrong?
<abralek>Hi glib 2.62 deprecated GTimeVal and now I cannot build javafx-graphics anymore
<abralek>I see that there is no other version of glib.
<abralek>What would be the right way to overcome this? Introduce public/private version of glib?
<TZander>abralek: the most logical way to fix this is talk to the javafx community to fix their code to work with the latest glib
<TZander>there might even be a patch somewhere that you can copy into the guix repo
<abralek>TZander: I agree, but this is 8u so I don't think it will happen
<abralek>actually build failes in gtk+2.24 (
<abralek>So if there going to be more gtk 2 release, everything should be fine
<rekado_>pinoaffe: is GUIX_LOCPATH set and does it point to a valid location providing valid locales for the effective version of Guix?
<pinoaffe>rekado_: GUIX_LOCPATH is /home/pino/.guix-profile/lib/local , I think that should be correct
<pinoaffe>srandom: ah yes, that was a pasting error of mine, it's actually set to /home/pino/.guix-profile/lib/locale
<lprndn>Hello Guix!
<sneek>lprndn, you have 3 messages!
<sneek>lprndn, ryanprior says: I'm definitely happy to work together on packaging elementary-related software. You can see my latest work here:
<sneek>lprndn, ryanprior says: so far I've got Granite, Appstream, Sideload, and Pantheon Terminal packaged. Of those, only Granite and Sideload are actually tested and working.
<sneek>lprndn, ryanprior says: I got my act together to submit an initial patch series for my Pantheon work. Take a look and tell me what you think:
<rekado_>I’m never quite sure of the semantics of sudo. Do you get the same error when using “sudo -E …”?
<PotentialUser-87>Hi, I've tried to install guix using the install script, and there is a missing file:
<PotentialUser-87>I'm on ubuntu systemd, x86_64
<PotentialUser-87>Could someone kindly share the content of this file?
<pinoaffe>rekado_: doesn't seem like it, thanks!
<idoesnot[m]>> Im trying to write a package definition for a package which only installs auxillary data for another package. The primary package needs to have environment variables defined to tell it where to find the auxillary files which I'm assuming can be done using the (search-paths) part of the package definition. However, when I write a package definition like this and install it with "guix package --install-from-file=foo.scm",
<idoesnot[m]>the environment variable I specified does not show up when using "guix package --search-paths". Is this something particular to the way I installed the package or is it something I've done wrong in the package definition?
<idoesnot[m]>Doing a bump of this question from earlier today in hopes that someone is online now that might know. Hope this is OK!
<rekado_>idoesnot[m]: I suppose it would help to see the package definition.
<mbakke>idoesnot: I think you need native-search-paths instead of search-paths
<rekado_>PotentialUser-87: there’s no such file. What tells you that the file is missing?
<PotentialUser-87>I think I got it:
<PotentialUser-87>If someone could just say what @storedir should be.
<mbakke>PotentialUser-87: this is because the Guix downloaded by the installer script does not yet contain that file
<mbakke>PotentialUser-87: can you file a bug report?
<mbakke>in the mean time, you can grab it from here, but you will need to do a few manual adjustments:
<mbakke>@storedir@ should most likely be /gnu/store
<PotentialUser-87>mbakke: thanks will try that.
<idoesnot[m]><rekado_ "idoesnot: I suppose it would hel"> Sure,
<idoesnot[m]><mbakke "idoesnot: I think you need nativ"> I thought I tried this but Ill try it again in case I missed it!
<PotentialUser-87>mbakke: submitted issue report.
<lprndn>sneek: later tell ryanprior: Nice! Most our work doesn't overlap. :) I'll look into that.
<lprndn>sneek: later tell ryanprior: On my side, I had a mostly working pantheon desktop on gnome 3.30. I'm updating everything I had. For now: plank, gala, wingpanel + most indicators, switchboard + most plugs, themes. It needs checking and testing. I also need to update files, calendar and screenshot and figure out what to do with some services...
<mbakke>PotentialUser-87: great, thanks
<srandom>I think it should be added to the manual page:
<srandom>substitute url address is not allowed to add '/' at the end
<kamil_>mbakke, have you received my message?
<mbakke>kamil_: what message? :)
<kamil_>Oh. That's odd. I sent a private message to you here on IRC. IRC is too confusing to me hahah
<kamil_>mbakke: This is the message: I asked you how to check the currently installed version of the kernel for CVE's, and how to use an older version of the kernel using a inferior. I successfully configured the inferior, pining the kernel package to an older commit, using examples in the Guix official manual and your instructions,where you told me to "(define kernel (lookup-inferior-packages ...))", as a point of reference. You also told me h
<kamil_>ow to scan the kernel package for CVEs, but I've not been able to get it right so far. You told me this: "you may also be able to call (@ (guix lint) check-vulnerabilities) directly in your system configuration, i.e. (define kernel (lookup-inferior-packages ...)) (check-vulnerabilities kernel)". My question is where in my config do I call (check-vulnerabilities kernel)?
<mbakke>kamil_: just after the channel definition should do
<mbakke>not sure it will work though :-)
<bricewge>lprndn: The screenshot in looks a lot like the recent issue with fonts is lightdm. WDYT?
<mbakke>bah, I think we need to graft non-bitmap fonts onto 'gs-fonts'
<bricewge>Before that issue with lightdm fonts, font-gnu-freefont wasn't installed in my system but the issue wasn't present
<bricewge>mbakke: What happened with that package (gs-fonts)?
<mbakke>bricewge: nothing, but Pango (the font rendering library used by almost everything) decided to no longer support bitmap fonts
<mbakke>gs-fonts are the only fonts available by default in build processes by virtue of being hard-coded into fontconfig, but are bitmap fonts and thus invisible to Pango users...
<alextee[m]><abralek "Hi glib 2.62 deprecated GTimeVal"> this also happens with gtksourceview. it's an upstream problem
<alextee[m]>actually it doesnt fail to build, just seeing the deprecation warnings
<alextee[m]>liberdiko: i sent a bug report about my fonts issue:
<bricewge>mbakke: Thank you! It makes sense know. Yestday lfam said such issues was pango related, but I didn't understood why: I couldn't find pango in lightdm-gtk-greeter's dependency graph
<bricewge>alextee: Yes, it was your issue I was linking to :)
<alextee[m]>oh just looking at mbakke 's link now
<alextee[m]>oh just saw that too lol sorry, slow
<alextee[m]>i thought having a dependency on gtk automatically means pango gets pulled, weird
<alextee[m]>how do you generate that dependency graph? i'll check
<jonsger>lol. when you try to fix a bug, you have no success but you fix another bug by-the-way :P
<bricewge>alextee: You are right, pango is present in the graph.
<bricewge>I was using the fancy xdot, I was probably misusing it but grep found it
<bricewge>'./pre-inst-env guix graph lightdm-gtk-greeter' =>
<alextee[m]>yeah pango is there
<abralek>alextee[m]: Indeed. I used -Wno-deprecated-declarations as well
<alextee[m]>abralek: 👍️
<alextee[m]>I asked in GTK channel about the font issue, and someone suggested that it could be abroken font-cache issue
<alextee[m]>"maybe the host fontconfig uses different cache structure than the packed one?"
<alextee[m]>"clearing the fc-cache would be the first thing I would try"
<abralek>alextee[m]: Well, yes, but I am trying to build jfx 8u with gtk2 and gtk3 support. I doubt that it make sense to change anything there
<alextee[m]>i also notice that in icecat sometimes the font is broken (shows as monospace) and i have to run fc-cache to fix it, maybe it's related
<alextee[m]>does the guix pack thing contain a font cache at all? i thought that's not part of the profile, but something on the user's system
<nckx>Allo' Guix.
<nckx>calher: You don't need to be scared to run ‘guix system reconfigure’ (or any Guix command) in parallel with ‘guix install’. And while it's true that you can almost always safely ^C it, one major exception is while GRUB is installing at the very end. Guix is transactional, GRUB is not; don't interrupt it. But that phase takes mere seconds.
<roptat>hi guix!
<nckx>issues.guix has got all fancee since I last saw it. Nice job, rekado_!
<kamil_>mbakke, yes, there's a problem. "(check-vulnerabilities kernel-package)" leads to "guix/ling.scm: In procedure check-vulnerabilities: In procedure %package-replacement-real: Wrong type argument: #<inferior-package linux-libre@4.14.180 xxxxxxxx>"
<mbakke>kamil_: OK, I guess you'll have to use 'guix time-machine' for now. It seems the linter cannot work with inferior packages atm.
<kamil_>mbakke, What a shame. Inferiors are still a technology preview, though. Thanks for trying to help me. I must figure out another way in which I can fetch CVEs for the currently booted kernel then.
<kamil_>(in an automated fashion)
*janneke procastinated writing that doc update for at least 3 months, phew!
<janneke>*procrastinated -- and i should reconfigure to get aspell going
<leon>Hi there! =)
<rekado_>srandom: or we should normalize the URL before use. Could you please report this to
<leon>I need some help, I did a guix install icecat, and when I launch icecat, I get this:
<alextee[m]>yay! more people experiencing this \o/
<rekado_>leon: there are a bunch of open bug reports about this
<nckx>So it begins.
<alextee[m]>im pretty sure the guix installation/packing phase is missing an fc-cache call somewhere
<rekado_>does it work for you to install font-dejavu and run fc-cache -f?
<leon>I'll try this
<rekado_>here’s one of the bug reports:
<leon>OH! Thank you very much! Installing font-dejavu worked for me.
<mbakke>hm, the standard (remove (cut string-prefix? "--foo" <>) ,flags) trick does not seem to work when flags is a not-yet-evaluated (list ...)
<leon>I have another question. My distribution is Debian Buster (Debian 10, lastest), and my desktop environment is GNOME. Do you think there could be a way to integrate icecat to the apps icons of GNOME 3? I'd like to run icecat without having to launch it from the command line.
*alextee[m] is also interested ^
<alextee[m]>would you have to create a .desktop file and install it on the user's system?
<alextee[m]>(or copy and patch the one from the package's store)
<rekado_>icecat has a .desktop file AFAIK
<mbakke>leon: export XDG_DATA_DIRS="$HOME/.guix-profile/share:$XDG_DATA_DIRS"
<rekado_>I think it’s just a matter of augmenting the search path by setting an environment variable
<rekado_>not sure if that’s safe
<mbakke>leon: you'd need to do that in GNOME's environment somehow
<nckx>Can't you point ‘Debian’ (totally out of my depth here) at .guix-profile/share/whatever?
<nckx>Which is probably what rekado_ is suggesting.
<rekado_>and what mbakke wrote
<nckx>You wouldn't need to patch or copy anything.
<rekado_>XDG_DATA_DIRS is probably the right thing here
<alextee[m]>ah nice
<rekado_>though I must admit to being wary of the XDG_* variables
<nckx>rekado_: Your message about mbakke's message has arrived, mbakke's message not yet. It's one of those days! \o/
*nckx opens logs.guix next to their IRC channel so they can IRC.
<rekado_>they also include directories where binaries might be stored (e.g. plugins) and then your foreign distro programs would load incompatible binaries
<rekado_>I feel that this XDG_* mess (caused by running Guix on foreign distros) needs a good solution
<alextee[m]>hmm i didnt know XDG_* affected the library load paths
<mbakke>a safer approach (and terrible hack) could be to symlink ~/.guix-profile/share/applications to /usr/local/share/applications, and make sure /usr/local/share is on XDG_DATA_DIRS
<mbakke>as the .desktop files reside below applications/
<rekado_>I wish we had bug reports for the problems that arise from setting XDG_* vars
<leon>Oh! That's awesome! It works now. I symlinked ~/.guix-profile/share/applications to /usr/local/share/applications, and now it works
<leon>Thanks a lot for all your help guys! =) That certainly was efficient.
<mbakke>leon: awesome, glad it worked :-)
<leon>But I also did the export XDG_DATA_DIRS="$HOME/.guix-profile/share:$XDG_DATA_DIRS" before, but that alone didn't work. So Does it work now because I did both this and the symlink? Or was it just the symlink? I have other machines elsewhere where I'm going to want to do it too.
<mbakke>leon: I don't think GNOME reads your local .bashrc or wherever you did that export.
<mbakke>you could check /proc/$(pidof gnome-shell)/environ to verify, I think
<leon>OK, I'll try this now
<leon>the XDG_DATA_DIRS is present in environ of gome-shell
<alextee[m]>where is the fc-cache output inside a guix-pack'ed package?
<alextee[m]>i mean, when i run this on another system, where does it look for the font cache?
<alextee[m]>i just tried fc-cache -rfv on the "gnu" directory but i still get font problems
<alextee[m]>maybe that's for the host system though, maybe i need to do edit the font cache inside "gnu" somehow?
<leon>Well, thanks a lot you people! Have a nice day. ;)
<nckx>alextee[m]: There shouldn't be any font cache files in /gnu. Packages don't contain font caches.
<belgin>how do I look for a package by the files it contains?
<nckx>belgin: That's not implemented yet.
<alextee[m]>nckx: so when i run a program from the guix packe'd package on another distro, it uses the host system's font cache?
<janneke>belgin: you mean like, readlink -f $(type -p ls)
<nckx>belgin: Well, you can ‘ls /gnu/store/*/bin/foo’ if you already *have* the package.
<nckx>Other (slow) hacks like that.
<belgin>i don't
<belgin>i'm looking for the package that contians glx.h
<nckx>But no ‘I want Xcmd, what should I install’.
<belgin>oh well, thank you for your help
<nckx>belgin: mesa, probably.
<nckx>On my system it's present in mesa, libepoxy, libxcb & sdl-union.
<belgin>ok thanks
<nckx>Mesa is probably what you really want.
<belgin>i installed it, but gcc-toolchain still can't find it
<nckx>alextee[m]: That is my understanding, but I'm not a foreign fellow.
<nckx>belgin: Well, which glx.h is it? GL/glx.h?
<alextee[m]>oh hmm, then it should work when i do `fc-cache -fv gnu`. maybe it's a different problem then
<pkill9>I'm adding a bootoader menu entry but I'm getting this build error, what does this mean? ice-9/boot-9.scm:1515:18: object is not an exception of the right type #<&gexp-input-error input: #<procedure base-initrd (file-systems #:key linux linux-modules
<pkill9>mapped-devices keyboard-layout qemu-networking? volatile-root? extra-modules on-error)>> #<record-type &package-input-error>
<alextee[m]>i can see that it properly picks up the fonts in gnu and caches them
<nckx>I'm installing gcc-toolchain but there's another thing I never use.
<alextee[m]>fc-list even shows the fonts in gnu!
<pkill9>I've set initrd to base-initrd
<alextee[m]>oh wait nvm it doesnt
<nckx>alextee[m]: And fc-match finds them (should, if fc-list does).
<nckx>That was a question but that doesn't matter now.
<alextee[m]>so it finds them and caches them, but it can't list them. weird
*alextee[m] posted a file: Screenshot from 2020-05-17 14-39-20.png (39KB) < >
<rekado_>pkill9: it’s picky. It expected an error, but not *that* kind of error!
<pkill9>oh haha
<belgin>nckx, i found it here /gnu/store/k7d62k2bmf97fw4ym3q26mygs2yxy9k5-profile/include/GL/glx.h, but apparently the toolchain can't see it
<abralek>OK, I gave another shot and have sent a cover email to guix-patches as it stated in the docs.
<rekado_>pkill9: I guess it wants to report an error to you, but didn’t expect that condition type, so you get an extra error on top of your actual error
<nckx>belgin: Yeah, I'm in the same spot.
<nckx>It's in mesa but not in the profile ($C_INCLUDE_PATH).
<rekado_>pkill9: I’m sure I sound like a corrupt MP3, but could you please send a bug report to Just what you did and what error you got?
<abralek>So now my question is do I need to add --in-reply-to=<ID> --no-thread to my git send-email command?
<pkill9>ok sure
*rekado_ builds a new version of guile-for-guile-emacs
<rekado_>pkill9: thanks!
<nckx>belgin: And that's because ‘guix environment --pure gcc-toolchain mesa’'s include/GL directory points to xorgproto/include/GL, not mesa's :-/
<rndd>hi everyone! i tried to install guix on my old laptop. after installation there are many errors and PC don't boot os. last error on display is "GC Warning: pthread_getattr_np or pthread_attr_getstack failed for maint thread". is it possible to fix?
<nckx>That's probably logical to someone who knows why but not to little old me.
<nckx>rndd: Do you have more output? That last line isn't relevant (happens to everyone).
<nckx>A screenshot is fine 's long as it's on a freedom-respecting site (no imgur please).
<rndd>nckx: for example?
<nckx>Well, that's the problem, isn't it.
<nckx>No way you can send the entire output so we can let our spider senses tingle?
<alextee[m]>aha!!! almost all of the .ttf fonts i have in gnu are symlinks, and they all point to /gnu/store (in the root!) and they don't exist!
<nckx>rndd: If I know what to look for I'd have solved your problem.
<nckx>alextee[m]: Yaywhut.
<rndd>nckx: nope, it is on another machine that refuses to boot
<rndd>nckx: dont have any tool to capture screen
<rndd>nckx: except camera on smartphone
<nckx>Right. Anything, say, ‘no such file or directory’-like?
<nckx>rndd: Uploading a photo would be fine.
<rndd>nckx: where i can upload photo>
<alextee[m]>how do you traverse all symlinks for a file?
<alextee[m]>ls -l just does 1 step
<nckx>readlink -m
<nckx>rndd: maybe? Never used them myself but libre-friendly.
*nckx has to go.
<alextee[m]>i found the issue
<alextee[m]>all the .ttf's are missing!
<alextee[m]>they are symlinked to /gnu/store...
<alextee[m]>which doesnt exist on a foreign distro
<alextee[m]>( i used -RR during guix pack)
<rekado_>are you using the new fakechroot backend?
<alextee[m]>oh those in the profile at least
<alextee[m]>rekado_: i believe so, i upgraded recently and i saw a news thing about it
*alextee[m] generates a debianpaste
<raghavgururajan>Hello Guix!
<alextee[m]>well i tried copying "gnu" to / and the symlinks work now but fc-list still doesn't find them (after fc-cache -f /gnu) and i still get boxes
<jonsger>I had h264 working with proper audio for some minutes in icecat. But I don't know what I changed in about:config :(
<rekado_>alextee[m]: that’s not unexpected. You are running find outside of the wrapper that bends the paths.
<rndd>nckx: any ideas 0_0?
<mbakke>rndd: what hardware is that?
<mbakke>apteryx: are the inkscape patches good to go? :-)
<rndd>mbakke: what info exactly you need?
<mbakke>rndd: just curious about the brand and model of the computer
<mbakke>the kernel says it's a BIOS issue, so not sure if we can work around it
<mbakke>google probably knows a thing or two about it
<rndd>mbakke: i tried trisquel and it works. afterall it is HP 15-db0146ur (Product code: 4MJ44EA)
<mbakke>rndd: does the boot stop completely on that last line (pthread_getattr_np failed)?
<mbakke>I think that specific warning is harmless, and if there is a kernel issue there should be a panic.
<nckx>rndd: I was away, as noted. It's possible that all those ACPI errors are non-fatal. I don't think we can say. Can you scroll back shift+page up?
<nckx>Your system doesn't look locked up or panicked, that's the (slightly) odd thing.
<rndd>nckx: nope i can't. or don't know how to (tried pgup and pgdn
<nckx>With Shift down?
<rndd>nckx: nope
<nckx>rndd: How standard is your system.scm? Very custom or installer defaults? A desperate hack you could try to get slightly more output is booting with ‘quiet fbcon=font:VGA8x8’ on the kernel command line.
<rndd>nckx: default one
<rndd>nckx: how to boot with ‘quiet fbcon=font:VGA8x8’ on
<rndd> the kernel command line.
<nckx>Press ‘e’ at the GRUB menu, add that text to the end of the ‘linux …’ line, hit C-x to boot.
<nckx>C means ‘Control’ here.
<nckx>So this is a Ryzen system? I think they're known to be troublesome. Don't know why.
<nckx>(Well, the buggy as hell BIOS might be a factor. -_-)
<rndd>nckx: well, i booted with ‘quiet fbcon=font:VGA8x8’ and nothing really changed in output
<nckx>Hmkay. It should've shrunk your font but never mind. It was pretty desperate. The problem is your screenshot full of text is telling us close to nothing useful, so it's going to be hard to suggest solutions.
<nckx>No search results on the Web?
<rndd>nckx: well, i can boot tqisquel and cannot guix. is it useful info?
<mbakke>rndd: does the boot stop completely on that last line (pthread_getattr_np failed)? how long did you wait?
<rndd>mbakke: ye, it stops on that last line. Around 30 minutes
<nckx>rndd: It gives hope that this can be fixed, yes. It doesn't immediately point us towards a solution. It might be useful to compare Trisquel's dmesg to the screenshot above. If it doesn't contain the same ACPI errors we know our kernel is doing something wrong/strict. If it does, we know they're not fatal on their own.
<rndd>nckx: should i send dmesg output here?
<rndd>i mean when i will install trisquel?
<nckx>If you do, use a pastebin (such as Make sure it starts at the beginning (the first line should be ‘Linux version 5.x…’), otherwise you'll have to get the full dmesg from $wherever_trisquel_saves_logs.
<nckx>Or send it to the mailing list. But the comparison is something you can do yourself, and so is checking for obvious differences.
<nckx>I think there are more AMD users on the list 😛
<rndd>nckx: another question. what i as newbe should learn in next few years to be able to solve such problems?
<rndd>well, maybe "few" is wrong word
<nckx>rndd: Honestly, just spend a lot of time breaking things and fixing them…
<erikg>I've run into a pretty strange problem with gcc 10.1.0 installed via guix:
<nckx>Ideally while someone pays you for the privilege, but it's a fun hobby too.
<erikg>basically i can't apply std::stoull to a std::string
<erikg>it requires a std::wstring
<erikg>nothing in the gcc release notes suggests that this should have changed
<nckx>Is there a non-Guix GCC 10.1.0 to compare?
<erikg>nckx: I'd have to build it from scratch
<erikg>so I thought I'd ask to see if this was a known change or issue
<nckx>erikg: You're better off asking in #gcc first, if that channel is alive.
<erikg>nckx: fair, I thought I'd nudge here in case it was a guix
<erikg>specific thing
<erikg>seems strange to have this land in a release if it's actually a bug
<nckx>I agree, but it would be a pretty convoluted Guix bug that would cause that subtle difference in behaviour without breaking things completely.
<nckx>Plus it would be more motivating to go bug-hunting in Guix if upstream actually says ‘nope, our gcc binaries don't do that’.
<erikg>of course
<nckx>We shall see!
<nckx>Who will win the weird bug of the week.
<erikg>gcc has locked down their bug reporting system behind a manual authorization of new users
<erikg>which is concerning
<erikg>but i can understand why
<mbakke>nckx: please use merges instead of cherry-picks to core-updates
<mbakke>also, we already have a newer cups and cups-filters on 'staging'
<nckx>It is a merge.
<nckx>And a core-updates package :-/
<mbakke>nckx: does not look like a merge:
<mbakke>merges look like this:
<mbakke>CUPS did not add much to the rebuild counts of the other patches on 'staging' so I decided to "throw it in there"
<nckx>I know what merges are, I used ‘git merge upstream/master’.
<nckx>How does
<nckx>commit ec7fc4aecd8a75ec3e38903a3212480d853ccd66
<nckx>Merge: 8d33f263c1 ebbf915422
<nckx>Author: Tobias Geerinckx-Rice <me@saibot.rg>
<nckx>Date: Sun May 17 14:44:24 2020 +0200
<nckx>Merge remote-tracking branch 'upstream/master' into core-updates.
<nckx>turn into that shitshow.
<nckx>Thanks for the invisible newlines too, Hexchat 👍
<mbakke>nckx: not sure, but that commit is not on core-updates
<nckx>mbakke: Does rebase destroy merge commits?
<mbakke>nckx: I think so.
<nckx>mbakke: How do you handle changes on c-u while you're working on them?
<mbakke>nckx: I used to abort and do a new merge (solving the same conflicts anew) for a long time, until I discovered 'git rerere'
<mbakke>nckx: though 'Merge: 8d33f263c1 ebbf915422' was the tip of the core-updates branch, no?
<nckx>I'm not sure, I'm doing reflog autopsy at this point.
<nckx>Just when you mentioned ‘rerere’ I was reading ‘rere is not the solution, as you still have to resolve merges manually the first time across.’ although this was a conflict-free merge, so maybe not relevant.
<mbakke>right, 'git rerere' is only used to save merge resolutions and irrelevant when there are no conflicts
<mbakke>even with git rerere you need to 'git merge --abort' and run a new 'git merge' if the branch has moved
<nckx>I've never used git rerere so I'll keep that for another time.
<efraim>I'll wait a bit before pushing my core updates packages
<nckx>efraim, mbakke: What needs to be done?
<nckx>I dunno, efraim seemed to be implying something needed to happen. I wouldn't know what.
<nckx>mbakke: So it seems I should have used ‘--rebase-merges’, is that correct?
<nckx>(Feel free to tell me you're not my git senpai & bugger off.)
<rekado_>good news: the guile-emacs build no longer times out with my rebased and patched guile-for-guile-emacs
<mbakke>nckx: no idea what that flag does :P
<apteryx>rekado_: super cool!
<rekado_>the bad news: it fails.
<rekado_>but that’s expected
<apteryx>I'm happy that someone picked it up
<rekado_>guix-emacs.el requires seq.el, which is not available in Emacs 24.4
<rekado_>so I’m just disabling that.
<rekado_>next step is to rebase Guile Emacs on top of Emacs 26.
<rekado_>(not gonna touch 27 because of the new dumping mechanism)
<nckx>mbakke: Fair enough. Sorry for the rebase mess, not sorry for core-updates CUPS, although I will check staging for sneakiness from now on 😉
<efraim>i figured you might want to ninja-delete core-updates and try again with a merge :)
<nckx>I would if I thought I could get away with it.
<mbakke>nckx: dw, not blaming you for anything, just thought you should know :-)
<mbakke>I'd recommend against rebasing merges if possible as the REBASING MERGES section of git-rebase(1) reads like an instruction manual for shooting your own foot :P
<nckx>mbakke: No, thanks. I hadn't got the mail blast when you did so I was clueless, having used ‘git merge’ as always.
<nckx>Yeah, I was just about to use the word ‘foot’ here. Cool, rebases destroy even non-conflict merges, thanks for letting me know this way git. Really appreciated. You rock. Next time also send a mail to my mother calling a stupidhead thanks.
<rekado_>bah, I have guile-emacs that builds and starts, but when I eval (display "hello") after entering scheme-interaction-mode I get a crash
<rekado_>ERROR: Wrong type to apply: #f
<janneke>...and it doesn't say which symbol happens to be #f ;-)
<janneke>you upgraded guile too, right?
<rekado_>yes, I rebased wip-elisp on top of stable-2.2
<rekado_>the last line of the backtrace is (_ misc-error "public-lookup" "No variable bound to ~s?" ?)
<rekado_>will investigate later
<janneke>great to see you working on this, anyway!
<rekado_>the example at does work though
<janneke>nice, that's already more than "something"!
<efraim>wee! upstreamed some ppc patches and rebased wip-ppc
<janneke>efraim: that's up which stream?
<alextee[m]>is there a guide/instructions/example packages for cross compiling (mingw) ?
<boeg>Save for just trying it out, how would I go about finding out if my hardware is supported by guix? Im running arch Linux on a MacBook Pro 12,1 and it works great. Should guix have the same hardware support available in the package archive?
<alextee[m]>ideally i would be making a mingw package for each of my dependencies?
<janneke>efraim: ah, up our own stream -- that's great!
<efraim>ah, yeah. missed your first message
<efraim>right now full upstreaming isn't possible with glibc-2.31, there's some sort of alignment errors
<efraim>locally I downgraded to 2.30, i'll re-create bootstrap binaries with 2.32
<ryanprior>I've got a patch series I'm ready to send with some new Python and Emacs packages =D
<sneek>Welcome back ryanprior, you have 2 messages!
<sneek>ryanprior, lprndn says: Nice! Most our work doesn't overlap. :) I'll look into that.
<sneek>ryanprior, lprndn says: On my side, I had a mostly working pantheon desktop on gnome 3.30. I'm updating everything I had. For now: plank, gala, wingpanel + most indicators, switchboard + most plugs, themes. It needs checking and testing. I also need to update files, calendar and screenshot and figure out what to do with some services...
<ryanprior>Anybody know how to use Emacs to send patches?
<janneke>efraim: yeah, saw your FIXME -- that's terrible
<ryanprior>What I've been doing is writing emails and attaching the patches as files, but since the patches are formatted as emails, that feels kinda extra?
<ryanprior>I assume there's a way to just send a patch as an email, like that's it that's the email.
<jetomit>ryanprior: there's git send-email
<jetomit>which mostly does the right thing™
<ryanprior>okay I haven't heard of that, searching now
<jetomit>this covers basic usage:
<bricewge>Too bad that 'git send-email' can't be used from magit
<abralek>Hm, Sending patches over the email is a new process for me. I did install send-email. But most of my question relates to the way to send them. settings for format-patch, writing a cover letter and send series
<abralek>Ideally I think I need to write a cover letter with patches and somehow make cover email available in guix-devel list as well as in guix patches
<erikg>nckx: no word from #gcc, seems like a dead channel
<abralek>Maybe it would be helpfull to have git send-email/format configuration topic in the docs?
<erikg>searching on the web doesn't explain why the string to integer conversion functions in gcc 10.1.0 now require a wstring
<efraim>I normally go with 'git format-patch -10 -o ~/my-patch-set' (-10 for 10 patches)
<abralek>efraim: OK, but what about chainreplyto/ thread.
<efraim>git send-mail --no-thread -n
<erikg>efraim: curious if you can reproduce this, it's blocking build of a few packages in
<efraim>'-n' because I can never get the reply/thread stuff right
<ryanprior>I believe I have sent an email using git. Amazing new frontiers.
<ryanprior>Magit Forge has "email as a pseudo-forge" on its roadmap liberdiko
<boeg>with guix, and possibly the nonguix channel, should I be able to have the same hardware support on my device as I have with arch without manually compiling drivers and stuff?
<efraim>erikg: you installed gcc-toolchain@10 and not just gcc@10, right?
<ryanprior>boeg: upstream guix does not recommend or package any nonfree drivers, so our substitute servers don't have them either.
<ryanprior>If you install a nonfree channel you'd have to also trust a nonfree substitute server in order to avoid compiling them.
<ryanprior>In any case the compilation will not be manual, guix handles all compilation manually using its build daemon.
<ryanprior>err handles it automatically**
<abralek>efraim: Can you clarify what is going to happen if I send a series of patches using thread to guix-patches? So all emails except the first one will have the first one in theirs references. but will debbugs create one ticket only?
<abralek>I was looking for some debbugs testings package/server but failed
<ryanprior>abralek: you need to send the first patch (or cover letter) first and wait for a response from debbugs, which will give you a ticket # and a reply address for that ticket.
<ryanprior>abralek: then you can send the rest of the patches to the given reply address
<bricewge>ryanprior Yes but unfortunately tarsius don't use email with git
<ryanprior>abralek: I'm not sure what you mean by "using thread" so sorry if I'm not being helpful =D
<efraim>git send-email 0000-cover-letter
<efraim>and then you get a response email giving you a patch number
<efraim>that patch number is the one you send the rest of the patches to
<boeg>ryanprior: i'm not sure if I need any nonfree drivers actually. Haven't installed any manually on this arch setup, just installed arch and turned on the correct wifi module (brcmfmac) and everything else worked out of the box, and as far as I can see, the brcmfmac driver is open source
<efraim>as far as threaded vs not threaded, we're not particular, we can figure it out :)
<abralek>ryanprior: That is what I did today. I sent series of patches and my first patch became a parent. But would it be better to set --nothread and make My cover email as a parent?
<abralek>efraim: I also would like to send cover email to guix-devel and have XXX in a cc
<abralek>So I am guessing there is no a single command to do everything in one go
<ryanprior>boeg: arch includes some non-free blobs in the kernel already iirc. check out which is arch with all known non-free components removed.
<TZander>abralek: you can just do 'git format-patch' and generate a list of files which you attach to an email in your email app.
<ryanprior>If Parabola works on your system then Guix System upstream likely will too; and the inverse.
<boeg>ryanprior: i guess i can just load guix via a live usb, and if stuff works there, it works, right?
<efraim>erikg: it looks like I got the same output
<abralek>TZander: yes, but afaik attachments is bad in such a process. Plus how patchworks will process this? How to send an updates in that case.
<ryanprior>boeg: I've never tried Guix System as a liveusb, I know there's a USB installer image but I don't know if it can boot a "live" environment? Anybody else know about that?
<efraim>erikg: I also go the same output with gcc-9
<TZander>abralek: as said above by someone else, you get an email confirming your bug number and you can send updates (any number of attachments) to that address. And thus they become available on that bug report.
<boeg>ryanprior: oh, i just assumed so
<boeg>ryanprior: like when i install arch, i get a live environment where i can connect to wifi, install packages, format drive and so on
<boeg>which i have already scripted, so its just a bash script i run that does it all, hence why im interested in guix
<abralek>TZander: yes, I know these steps. But check out this issue It is funny =)
<ryanprior>boeg: yes that's a desirable thing for sure!
<ryanprior>Parabola has a live image for sure.
<abralek>TZander: If you an emacs + debbugs user checkout the ticket as well. Also funny =)
<TZander>I don't use emacs, just KDE.
<boeg>but using the guix installer, does the installer not connect either via wifi or ethernet to download stuff? Because i mean if that is the case, then i can see if wifi is supported that way, ryanprior
<boeg>or are all packages includes on the installer?
<efraim>erikg: it also failed with gcc-toolchain@8 and only started working with gcc-toolchain@7
<abralek>I take advantage of framapic and put a screen here
<boeg>it looks like it does, so i think im just gonna try it
<TZander>boeg: guix uses a debian-like installer AFAIK
<boeg>yeah, i can see in the guided installer guide
<boeg>TZander: ^
<efraim>erikg: clang++ from clang-toolchain@10 worked
<abralek>Is it OK to send yet another series of patches, which depends on other not-yet-merged series?
<abralek>Don't know how the build will work here
<ryanprior>abralek: you probably need to wait for the first series to land.
<abralek>ryanprior: thanks
<ryanprior>abralek: I maintain a repo for my packages so that I can use them and link people to them before they're ready to send to the ml
<kamil_>What are your recommendations of emacs plugins available in the repository? I already have an eye on emacs-evil because I'm used to Vi editors.
<kamil_>I need a few basic ones that simplify writing GNU Guile code
<boeg>is it correctly understood that I can make a guile scheme "config file" and when that has been made, the only thing I need to set up a device with this config is to (i assume) manually format disks and install a "base system" and set up and user and then just apply the config and the rest of the system - drivers and locale and everything is - is set up as to my configuration?
<anadon>I'm having trouble getting `git send-email` configured just right. Coulld someone decipher the following for me?
<anadon>STARTTLS failed! SSL connect attempt failed error:1416F086:SSL routines:tls_process_server_certificate:certificate verify failed at /gnu/store/js8f0w00yp5mv4jkcx2nkzi10wsrjb2x-git-2.26.2-send-email/libexec/git-core/.git-send-email-real line 1548.
<anadon>nss-certs and libressl are both installed.
<apteryx>interesting; make check-system TESTS=installed-os -> guix system: error: copy-file: Success: "/mnt/gnu/store/l6afn2v9hkgms9basdf941k63dm1d5hr-locale-2.29/2.29/ru_RU.utf8/LC_COLLATE"
<jlicht>hey guix!
<reepca>boeg: the only part you have to do manually is creating the partition you want it on. From there you can just 'guix system init config.scm <target-partition>' from any system with a functioning guix on it.
<ryanprior>sneek: later tell lprndn: do you want to send a pull request to my repo ( or create your own repo for your pantheon packages, so I can try them?
<sneek>Got it.
<reepca>that could be the bootable installer image, guix running on guix system, or guix running on a foreign distribution
<anadon>Found the issue, I had not set up environmental variables according to the X509 section in the manual.
<ryanprior>sneek: later tell lprndn: we should put together a system definition that contains all the pantheon stuff so we can start testing it in a VM and work out the configuration.
<sneek>Got it.
<boeg>reepca: what about root and user password? stored encrypted in the config?
<boeg>or, stored as a hash, rather*
<reepca>boeg: see section 8.5 "User Accounts" of the manual.
<boeg>reepca: thanks
<reepca>specifically the 'password' field
<boeg>reepca: yes, thank you very much
<jonsger>hm icedove doesnt fill in the mail address when opening a link from chromium
<apteryx>mailto protocol failure in the .desktop associated file?
<jonsger>apteryx: what do you mean
<apteryx>perhaps something to do with (x-scheme-handler).
<Blackbeard>I did a guix pull and a guix upgrade and now inkscape is full of weird symbols
<Blackbeard>There is no text in the menus
<Blackbeard>Just symbols
<rekado_>hmm, Guile Emacs works only when run inside GDB
<Blackbeard>I tried to change the font using lxappeareance but this program has the same problem
<jonsger>apteryx: I have this in place
<jonsger>oh savannah is down again
<ryanprior>My guix-installed software has been degrading day after day in terms of font rendering.
<ryanprior>My Emacs has started showing symbols instead of text in places
<ryanprior>I really need a thorough explainer of how one is supposed to get guix apps to render fonts appropriately, and ideally also of why that isn't something we can get workign out of the box (or how I can help contribute to making that happen)
<rekado_>it's a regression
<rekado_>but there’s not enough information to tell us what to do about it
<apteryx>ryanprior: for now a possible workaround seems to be to install font-dejavu
<bricewge>Blackbeard: See
<bricewge>It should fix the issue, I didn't tried it yet tho
<maddo>If I may ask, what is the font used in the Guix PDF docs?
<maddo>and in the refcard
<cbaines>maddo, I believe the source for the refcard can be found here
<cbaines>I think GentiumPlus is mentioned as the font...
<leoprikler>so we need to install fonts on system level now IIUC?
<bricewge>leoprikler: Not with mbakke patch since it adds font-dejavu to fontconfig
<leoprikler>okay, but what if I want to use a font different from dejavu?
<leoprikler>e.g. I have trouble displaying both Cantarell and Google Noto in Epiphany
<leoprikler>is that even a related bug?
<jackhill>I have font trouble in Epiphany as well, but I haven't investigated enough to say more than that. I had been thinking of it as a Epiphany-spcific problem, fonts in Icecat and other software seem fine (but I don't use many font-heavy programs).
<jackhill>I have all the font packages installed in my user's default profile. For Epiphany I had been wondering if the fonts wheren't getting into the bubblewrap container for the web worker processes.
<jackhill>Oh! I just tried gnome-font-viewer, and it doesn't show any fonts. I did in the past. hmmm…
<leoprikler>hmm, that may certainly be the case
<leoprikler>rekado_: yep, that's mine