IRC channel logs


back to list of logs

<nckx>(That's the MSDOS disklabel, yours are GPT types which also sound plausible.)
<ennoausberlin>lsblk gives me iso9660 vfat hfsplus iso9660
<ennoausberlin>I am sure I could boot and install .16 a few weeks ago on another pc.
<ennoausberlin>The qemu image runs fine. But I use a hdmi pc stick here
<ennoausberlin>Well, I try the 32bit image. Hope, it will work
<nckx>ennoausberlin: what does ‘file /dev/sdx*’ say? (I'm just grasping at straws here TBH.)
<nckx>ennoausberlin: This is on a Linux system, right?
<ennoausberlin>sda a1 a2 a3 a4 sdb
<nckx>Whoops. Sorry: file -s /dev/sda*.
<ennoausberlin>Ubuntu 18.04
<ennoausberlin>Okay. Thats a lot to type. I write on the phone right now
<nckx>ennoausberlin: Oh, never mind then. Here, it shows 2 partitions, one ext4 (label: "GuixSD_image") and the other FAT (label: "GNU-ESP ").
<ennoausberlin>sda is dos/mbr. sda1 data. sda2 dos mbr followed by fat. sda3 hfsplus. sda4 data. hfsplus is the root partition I guess.
<nckx>If yours is very different, something's off.
<nckx>Any labels/volume names?
<nckx>Oh, interesting, when I ‘losetup -fP guixsd-install-0.16.0.x86_64-linux.iso’ I *do* get your four partitions.
<nckx>Same types & all. No idea why the kernel treats my USB drive & image file differently but I guess there's nothing wrong with your drive...
<ennoausberlin>no. hfsplus modification time is 1970. unix time. Very strange
<nckx>ennoausberlin: That's time stamp ‘1’. Again, very common in Guix to create reproducible images :-)
<ennoausberlin>Ah. I see. My fault
<nckx>Oh, no. Just something you'll get used to quickly.
<nckx>Sounds like your USB drive is fine, and you're entering the fun world of ‘why won't my BIOS boot this’. Sorry.
<ennoausberlin>Can I install guixsd from within a running qemu image? The qemu probably does not see my disks, right?
<ZombieChicken>you can make qemu access raw drivers iirc
<ZombieChicken>er, raw drives
<nckx>Just make sure it's not your raw mounted root device if you like having data.
<ennoausberlin>Big question then. Will it boot the medium after install? I will try
<bitmapper>ahh, i'm having fun trying to learn how to write a package for guix
<ZombieChicken>last time I tried to do something like that, I think I dded the raw drive image to an raw img file (it was a flash drive I was installing to), installed onto that, then dded the resulting image back to the device once the install was done
<ZombieChicken>but if you're trying to install to a large disk, that certainly isn't practical
<nckx>There are some other fun/insane workarounds you can try, like booting a USB drive that *does* work on your PC (presumably Ubuntu), entering the GRUB command line, and seeing if you can ‘configfile (hdX,y)/boot/grub/grub.cfg’ the GuixSD grub.cfg...
<ennoausberlin>The disk isnt that big. Later I will have the problem, that the store grows bigger than the size of my usb stick/sd card. :)
<nckx>Or installing the Guix binary tarball to a running Debian/whatever live system booted from USB, then using *that* to guix system init...
<nckx>* may not contain traces of fun.
<nckx>I've done both.
<ZombieChicken>overwritting a working OS sounds like a really good way to break a system in special ways
<ennoausberlin>nckx: Tried that grub command line method before but just got a frozen pc :).
<ennoausberlin>I like these pc on a stick devices but sometimes the uefi stuff drives me crazy. Especially while travelling without fast internet
<nckx>ZombieChicken: Actually, that's the installation method used on some of the ARM build machines. It works. I just don't like the left-overs of the previous distro all over the place.
<ZombieChicken>yeah. LIke I said, break things in special ways. Might work from a minimal Gentoo install or something, but I could see that causing some really special problems
*nckx .oO This may explain why my ARM build machines are still offline.
<nckx>ZombieChicken: OpenSuSE or however you capitalise it this week works.
<ennoausberlin>Well, I am going to bed. This took too long. But thanks to all of you. Great community.
<bitmapper>ugh, i can't figure out how to extract a tar.gz with trivial-build-system
<nckx>ennoausberlin: I hope you find the motivation to continue your quest tomorrow! :-)
<ZombieChicken>I think I'd start with a minimal Gentoo install, grab the checksum for everything it has, and once the guix install is done, check the checksums versus what is on the disk and delete what is left over
<bitmapper>does anybody have any ideas?
<ZombieChicken>er, delete what matches
<ZombieChicken>gunzip it then tar xvf?
<bitmapper>in a derivation?
<bitmapper>i would if i knew where it was
<ZombieChicken>if your talking about building a package, then I can't help.
<bitmapper>it's the source
<ZombieChicken>guixsd is one of those distros for me that I'd love to use, but can't because I refuse to replace working hardware
<nckx>bitmapper: Something like net-base does?
<nckx>(Random grep for ‘trivial-build-system’ and ‘xvf’.)
<ennoausberlin>nckx: I already have another guix running. But I need one for experimenting.
<nckx>bitmapper: in gnu/packages/admin.scm.
<bitmapper>i found it with guix edit
<bitmapper>guix is a bit more confusing than nix to be honest
<bitmapper>from the packaging side
<nckx>Hm. As an ex-Nixer I can't agree, but it seems to be a very personal thing.
<lfam>bitmapper: Interesting, I feel the opposite way :) I think it's more a matter of what you learn first
<bitmapper>i learned guile first
<nckx>Except I learnt Nix first and still prefer Guile.
<bitmapper>i guess it's just the docs aren't complete rn
<bitmapper>never said i didn't prefer it
<nckx>This is a true.
<bitmapper>it's just hard to figure things out compared to nix
<lfam>Good thing we are here to help :)
<bitmapper>i'm trying to package up discord for a friend of mine :/
<ennoausberlin>I am coming from Smalltalk and I like scheme, but debugging scheme is something I will never master
<nckx>bitmapper: Also, the trivial-build-system isn't Guix's friendliest face. The build system is trivial, your code won't be.
<bitmapper>yeah, but i'm dealing with prop binaries, so the gnu make system won't really work i guess
<ennoausberlin>bitmapper: there is a cmake build system
<bitmapper>people just don't want to move away from discord
<bitmapper>and for some reason one of them thought it was a good idea to install a libre distro
<nckx>Hehe. (Well, that's *always* a good idea. ;-)
<Swedneck>perhaps they're like me and really really like the idea of guixsd aside from the libre-only bit
<bitmapper>i mean i'm like that but i don't really use any prop software anyway
<ZombieChicken>lfam: Any chance for help using the vanilla linux kernel instead of linux-libre?
<ZombieChicken>ennoausberlin: How is Smalltalk as a language? I've heard interesting things, but never seen it used in the wild
<lfam>ZombieChicken: The official Guix communication channels (the mailing lists, website, this IRC channel) aren't to be used to promote the use of software without a free license, sorry
<Swedneck>i think the only prop. software i use is games really, aside from firmware and drivers, which is really the big issue with libre-only
<ennoausberlin>I have my own stories about non working sound and network, but I managed it with an cheap usb audio adapter and an older huawei usb wifi repeater
<bitmapper>i just installed an atheros card into my laptop
<bitmapper>only thing i needed to replace
<bitmapper>heh i'm a bit of an oddity i guess
<bitmapper>i'd like to eventually get a librebooted x200
<ennoausberlin>ZombieChicken: Its an very good designed oo language like ruby plus one of the best development environments. I use spacemacs for scheme and I tried slime for common lisp. But I always miss my Smalltalk environment :/
<ZombieChicken>ennoausberlin: Yeah. It's sad, but the Lisp family has lost some debugging features over time
<ennoausberlin>Okay. Its late. Bye bye and thank you again
<mikadozero>Would there be any problem with moving config.scm into /home/non-root-user/.config/ ?
<bitmapper>why would you need to?
<mikadozero>Moving it from /etc/config/
<bitmapper>what's wrong with it being in in /etc/config/
<ng0>the location doesn't really matter
<mikadozero>bitmapper: I was thinking I wanted to keep most or all of my config files under version control with git in the users home directory.
<nckx>mikadozero: That's fine. The name is entirely arbitrary. I use /etc/guix/system.scm, for example.
<ng0>you could have this in /tree/guix/config.scm and it would still work as the same file. as nckx wrote.
<mikadozero>The reason I am thinking of having them together in a git repo was based an a suggestion of managing my configurations with guix by having a guix package for them. So I thought I could maybe create a guix package for all my configs.
<mikadozero>Based on one git repo of all my configs in the home directory.
<ng0>it's maybe a nice experiment, but having it in plain git removes an unnecessary dependency on guix.. for me having guix controling configs compared to just having make doing symlinks sounds not so practical
<ng0>but like, the location doesn't matter. I kept my guixsd configs in a system config repo
<ng0>for taler we keep it in a repo, and for gnunet as well.
<ZombieChicken>Wasn't there a plan where guixsd systems could talk to each other and confirm the checksum of packages on systems to help determine if a package was incorrect?
<ZombieChicken>also, can I include a package definition in a configuration file?
<nckx>ZombieChicken: Yes.
<ZombieChicken>Ah. Is rootless X an option in GuixSD atm?
<nckx>ZombieChicken: Working GDM in Guix System is only a few days old (if it's even been merged yet), and someone mentioned that as a prerequisite. So my guess'd be no, not yet.
*nckx knows nothing tho'.
<ZombieChicken>GDM == gnome desktop manager or something?
<ZombieChicken>I'm not sure what software your refering to
<ZombieChicken>that isn't a prereq unless someone is doing something really stupid
<nckx>Ah, optimism.
<ZombieChicken>didn't say something stupid wasn't being done, just that it gdm shouldn't be a requirement
<nckx>‘If true’ so yah who knows.
<ZombieChicken>considering I'm on Void Linux atm and I'm not using GDM and I'm using rootless X, I'll say that "if true" statement is false unless someone is trying to force Gnome down people's throats
<ZombieChicken>and I don't ahve gdm installed. Just checked to make sure
<nckx>ZombieChicken: As you tick both boxes ‘seems to care about it’ and ‘seems to know something about it’ , chances are you can make a real difference by helping out.
<nckx>Which DM do you use?
<ZombieChicken>startx -- vt1
<ZombieChicken>I honestly have no clue why people like DEs
***catonano_ is now known as catonano
*nckx equally has no clue why people still use VTs.
<ZombieChicken>in my case it's a lack of good alternatives and a safe place to start. If something breaks X, I'm not having to hope I can get to a VT to do something to fix things
<lfam>Guix has room for both camps
<nckx>Last time I broke X was changing my CRT's modelines in xfree86config.
<nckx>Good times.
<ZombieChicken>I've also yet to find a DM that works well that isn't tied to KDE or Gnome
<nckx>Does Guix provide startx though? ls is taking ages to tell me...
<ZombieChicken>it's usually in xinit for most distros I think
<ZombieChicken>and iirc, startx is just a shell script
<ZombieChicken>/usr/bin/startx: POSIX shell script, ASCII text executable
<nckx>Right, but one that, like, works.
<ZombieChicken>it is here at least
<ZombieChicken>It means you could likely grab one from somewhere else and get it to work
<nckx>I'm asking because xorg-start-command in (gnu services xorg) seems to write one, and that seems like a strange place to do it if the vanilla xinit version worked.
<lfam>I don't remember the specifics (search the mailing lists) but the vanilla startx did not work on GuixSD
<lfam>There are reasons™ that X / DE stuff is the way it is on Guix
<nckx>ls has returned! Hm. It's a (long) shell script.
<lfam>It's not that people are stupid or never tried the obvious things before...
<ZombieChicken>other than the weird path issues?
<nckx>Looks meh. Paths are patched though.
<lfam>I don't know, it warrants a search through the mailing list archives for sure
<ZombieChicken>yeah, but I don't know the issues other than the obvious "X isn't in /usr/bin"
<ZombieChicken>I should say $X
<ZombieChicken>Oh well. Going to see if I can get this vm working
<nckx>I wish I could say what lfam says so nicely without sounding a lot more cranky :-)
<lfam>This looks really promising:
<lfam>And even includes suggestions for further work :)
<nckx>‘We will run X server with user privileges’, too.
<lfam>It's really 2019
<ZombieChicken>yeah, that's a really nice link
<ZombieChicken>looks like it doesn't have to be installed by default, either
<ZombieChicken>which is even better
<lfam>We figured it out just in time to make the big switch to wayland ;)
<ZombieChicken>better late than never
<lfam>Yes, part of Guix is devolving power to unprivileged users, so that method sounds nice
<nckx>Noob question: Wayland is unpriv'd by default, right?
<bitmapper>i'm getting frusterated
<bitmapper>why am i getting "In procedure link: file exists"
<bitmapper>fixed it
<ZombieChicken>Is there a needed order to file-systems and mapped-devices in the system config?
<Elon_Satoshi>ZombieChicken: if there is, devices have to be mapped before they can be mounted to file systems
<ZombieChicken>yeah, but I wasn't sure if the config was smart enough to figure things out or not
<Elon_Satoshi>I don't know either. I put my mapped device before the filesystem definitions to be safe
<ZombieChicken>yeah, that's what I'm doing
<Elon_Satoshi>I just looked at the package for ledger, it looks complicated. How does one figure out all those cmake parameters and dependencies?
<reepca>Elon_Satoshi: trial and error, reading docs (where they exist), looking at how others have packaged it elsewhere, etc.
<Elon_Satoshi>how long does it take to package a program?
<Elon_Satoshi>Who packaged ledger? Do you think packaging hledger will be too difficul?
<vagrantc>really depends on the program
<vagrantc>if it's a "make && make install" or some other straightforward build system ... probably not too hard
<vagrantc>if things it needs to build aren't packaged, then you get into a rabbit hole like i did the last week of packaging a chain of dependencies
<Elon_Satoshi>tell me about it
<vagrantc>but quite a bit is packaged already
<vagrantc>and "guix import" can help quite a bit
<ZombieChicken>okay, so what is the syntax for (dependencies ...) in the file-systems section of the system config? I need to reference a mapped-device
<Elon_Satoshi>Are guix packages sometimes long because the program can be compiled simply with "make && make install", but the author decided to add a bunch of default make/cmake variables in order to easily edit them?
<Elon_Satoshi>For instance, ledger seems to be simple to install:
<vagrantc>or they get it mostly working with "make .." but then you have to manually tweak things ... or force it not to look in standard library paths ... or a handful of things
<vagrantc>hledger looks to be haskell, so it could definitely have a chain of depends ... not sure if there's a haskell importer
<Elon_Satoshi>As I said, ledger is easy to compile according to that but when I run `guix edit ledger` I see all kinds of scripting, cmake flags, and stuff
<Elon_Satoshi>Hmm. Maybe it's because this ./acprep file is meant to set things up for systems other than Guix
<Elon_Satoshi>This file also does a lot of complex dependency installing and cmake flags
***rekado_ is now known as rekado
<rekado>vagrantc: there is an importer for stackage and for hackage.
*vagrantc tried to dive into haskell, but never quite stuck
<rekado>Elon_Satoshi: we don’t use acprep but the default CMake build system for ledger.
<rekado>extra build phases were added to fix compatibility bugs, to build and install documentation, and to install the Emacs mode files.
<roptat>Hi guix!
<rekado>roptat: hello!
<rekado>Elon_Satoshi: some of these phases are no longer needed for version 3.1.2. (I’m updating ledger right now.)
<roptat>Hey, I received an email through a mailing list. They are looking for speakers on reproducible science. I think it's a good opportunity for a guix tblk, no?
<roptat>The email is in French, and I think they are looking for French-speaking speakers, should I forward it to guix-devel in case someone is interested?
<cbaines> has a red cross by e7e259a, but it seems to work find for me locally...
<efraim>Even after make clean && make?
<cbaines>yeah, I didn't spot anything in the output
<cbaines>the next build looks to have worked
<rekado>hmm, I can’t reach the git repo on Savannah.
<rekado>ah, there we go
<notnotdan>Hm, so `invoke' is basically the same thing as `system*' but it also checks that the return code is non-zero?
<jlicht>hey guix!
<notnotdan>hi jlicht
<notnotdan>Is lambda* defined in Guix or in Guile?
<tune>If I rolled back from generation 81 to 80 and then decided to try updating again, will that put me on generation 82 or overwrite 81? also if that new generation is also broken, will a rollback bring me to 80 again or back to 81?
<rekado>it would create a new generation and link that to 81
<rekado>the former generation 81 would still exist but its link would be gone.
<tune>interesting. okay
<civodul>Hello Guix!
<jlicht>hey civodul
<nckx>o/ civodul
<civodul>hey hey!
<ArneBab>notnotdan: in Guile, see ,a lambda*
<bitmapper>i don't really understand how to define dependancies in guix packages
<bitmapper>could someone help?
<rekado>bitmapper: dependencies must be declared as inputs.
<bitmapper>propagated inputs?
<roptat>depends on what you want to do
<roptat>native-inputs if they are used for building the package, inputs for dependencies like libraries
<rekado>bitmapper: there are inputs, native-inputs, and propagated-inputs
<roptat>propagated-inputs is for dependencies that need to be installed in the profile too, we try not to use it though as it can create conflicts easily with other packages
<roptat> has some better explanations on the difference between the three kinds of inputs
<civodul>rekado: could the manual on use the same CSS as the one on
<civodul>(speaking of which, i recently realized i have write access to that CSS, in case we want to make incremental improvements)
<rekado>I wondered about this too
<rekado>all I did was run “make html”; then I copied the thing over.
<rekado>where can I find the css?
<civodul>you'd need to run from Gnulib
<civodul>and that produces the complete manual/ directory
<rekado>I see
<civodul>it works well though :-)
<rekado>I’ll add it to my list; will look into it tonight
<civodul>we could have an mcron job for that
*rekado has to package a few Java things now :(
*civodul sympathizes
<roptat>rekado, maybe I've already packaged them in my repo? there are a lot of packages there
<civodul>at least there's quite a few awesome Java packagers now
<civodul>not like when you started Java packaging, rekado :-)
*civodul just pushed... "guix system delete-generations"
<civodul>fearlessly tested on the bare metal!
<rekado>roptat: unlikely. It’s called Juicer and is a “one-click system for analyzing loop-resolution Hi-C experiments”
<rekado>the “one-click” part is a lie
<rekado>even if you ignore all the Java stuff
<rekado>it’s a bunch of scripts that … frustratingly don’t do anything sensible without heavy editing.
<roptat>ok, if you need dependencies that are in my repo though, don't hesitate to copy :)
<rekado>roptat: will you push them from your repo to Guix upstream eventually?
<roptat>yes I will
<rekado>(or should I say “sweat” now? Not sure about the channel rules…)
<roptat>I'm still working on them and trying to make sense of the dependencies of everything...
<roptat>I could push josm and its dependencies, but it looks like the svn-fetch is still not completely reproducible
<roptat>I can get the same hash multiple times in a row, and then it changes after some time, stays stable for some time and changes again...
<rekado>can you see what’s different each time it changes?
<roptat>I suspect some file changed, but I could experiment a bit, yes
<roptat>it's a svn repo with external sub-repos
<roptat>I suspect it's fetching the right commit for josm's repo, but the latest for every other sub-repo
<bitmapper>does guix have libcxx/libc++
<nckx>bitmapper: The LLVM one? I don't think so.
<bitmapper>uh oh
<bitmapper>sudo: /run/current-system/profile/bin/sudo must be owned by uid 0 and have the setuid bit set
<nckx>bitmapper: Hm. /run/setuid-programs/sudo (the setuid version, natch) should come first in $PATH.
<nckx>Weird that it doesn't.
<rekado>(I thought I packaged the LLVM libcxx…?)
<bitmapper>nckx: was a screwed up path leftover from debugging
<bitmapper>it's been like that for a while
<bitmapper>weird i haven't needed to use sudo
<nckx>rekado: I can't find it in master. Looked for both libc++ (yay) and libcxx (boo).
<bitmapper>would it be part of clang-runtime?
<bitmapper>...and now no substitutes are working
<nckx>bitmapper: I also ran a quick (ha ha, lies) find over the store of my build box. I don't think we have it yet, at all.
<bitmapper>lookup error
<nckx>bitmapper: Were you the one who reported that same problem y'day? It doesn't sound Guix-related.
<bitmapper>my interent is just acting funny
<Elon_Satoshi>`guix build qemu-minimal@2.10.2` isn't working, how do I compile qemu-minimal-2.10.2?
<bitmapper>there's no libffmpeg?
<sisyphe_>hi #guix, when trying to cross build package I stumbled on "build system `python' does not support cross builds" but when looking at python-build-system.scm I cannot figure what part of its definition make it incompatible with cross build
<sisyphe_>How does guix even known which build system support cross build and which does not ?
<rekado>sisyphe_: take a look at guix/build-system/python.scm, the “lower” procedure.
<rekado>you’ll see that it check if target is specified. If it is “lower” returns #F and does not perform any work.
<Elon_Satoshi>Does anyone know how to build qemu-minimal-2.10.2? I wanna see why it freezes my computer up while compiling
<bavier>hello guix
<sisyphe_>thanks, found it
<Elon_Satoshi>hey bavier
<bavier>anyone use btrfs subvolumes on a guix system?
<Elon_Satoshi>bavier: How do I compile qemu-minimal-2.10.2 so I can see why it's causing my computer to freeze?
<notnotdan>Elon_Satoshi: try using `--verbosity=5` option?
<Elon_Satoshi>notnotdan: Thanks, but my problem is that Guix can't even find it
<rekado>can’t find it?
<Elon_Satoshi>$ guix build qemu-minimal@2.10.2
<rekado>how can it freeze your computer then?
<Elon_Satoshi>guix build: error: qemu-minimal: package not found for version 2.10.2
<rekado>when it builds it, it will print the derivation file
<rekado>try “guix build /gnu/store/….drv”
<rekado>the package is hidden.
<Elon_Satoshi>rekado: Wait a minute, where can I find the log for compiling qemu-minimal-2.10.2?
<nckx>Elon_Satoshi: It's hidden. You have to use (@@ (gnu packages foo) qemu-foo) on the command line.
<nckx>* Forgot -e. guix build -e '(@@ (gnu packages foo) qemu-foo)'.
<nckx>Something like that anyway.
<rekado>Elon_Satoshi: when Guix builds things it tells you what it builds.
<rekado>Elon_Satoshi: it will say something like “building derivation /gnu/store…drv”
<rekado>to build that same thing you can use “guix build …drv”
<Elon_Satoshi>Alright. I will try that if I have to. But, I just realized that somewhere there is already a build log
<Elon_Satoshi>...well I hope.
<bavier>Elon_Satoshi: I just got a substitute for qemu-minimal-2.10 from hydra, you're building with --no-substitutes?
<bavier>(in any case, it'd be nice to figure out what's going on)
<roptat>Elon_Satoshi, doesn't guix tell you about a log file when it fails to build qemu?
<Elon_Satoshi>I just noticed there are directories called /mnt/guix/tmp/guix-build-qemu-minimal-2.10.1.drv-0
<roptat>like "View build log at '/var/log/...drv.bz2'"?
<Elon_Satoshi>roptat: there's no chance those logs would go onto my hard drive via cow-store, is there?
<Elon_Satoshi>the logs disappear as soon as I reboot my computer
<raghavgururajan>1) When I installed the Guix System through "guix system init", I used the config "desktop.scm", which already installed gnome with networkmanager. But when I ran "guix package -i network-manager", it's installing like it never installed before. That is, instead of showing "following packages will be upgraded: network-manager", it's showing "following packages will be installed: network-manager". So packages mentioned in config file will not
<raghavgururajan>registered in profile even though they technically get installed during "guix system init"?. 2) I installed openvpn, network-manager, network-manager-applet and network-manager-openvpn. But I am not able to import openvpn profile via networkmanager applet. Openvpn option is not appearing in the networkmanager at all. What should I do?.
<bavier>raghavgururajan: system installed packages are separate from package installed to a user's profile
<roptat>raghavgururajan, for 1), guix actually has more than one profile: network-manager is installed in the system profile, but guix package only modifies your user's profile
<raghavgururajan>Ahhh! Now i get it. Thanks guys!
<raghavgururajan>Any solution to the 2nd problem please?
<roptat>I don't know, sorry :/
<raghavgururajan>I also have another problem. Let me know if you have any ideas.
<raghavgururajan>3) Bluetooth is working when I login via root user. But when I login as different user, there is no bluetooth option in networkmanager applet. If I go go bluetooth option in system settings, it says no bluetooth device found.
<raghavgururajan>*go to
<raghavgururajan>Is there any "bluetooth" group that I have to add the user to??
<nckx>I don't have Bluetooth, but from the manual (7.7.7): ‘Users need to be in the ‘lp’ group to access the D-Bus service.’
<nckx>(The service being bluetooth-service.)
<nckx>raghavgururajan: ^
<quiliro>Is there a way to know which packages are installed by guix system after installation?
<raghavgururajan>@quiliro: I have the same doubt. Also, can anyone tell me whether it is possible to install/remove packages as system instead of user?
<nckx>raghavgururajan: Yes, by removing them from PACKAGES in your system configuration.
<nckx>& reconfigurin'.
*Elon_Satoshi is running `guix system --verbosity=5 --no-bootloader init /mnt/guix/etc/config.scm /mnt/guix`
<Elon_Satoshi>to heckers with --no-substitutes
<isengaara-tomoko>I tried building gnu hello in the the wip-gcc7 branch, but every time gnu guile fails to build
<nckx>quiliro, raghavgururajan: Try: guix gc --requisites `guix system build /etc/guix/system.scm`.
<lfam>isengaara-tomoko: There is not a wip-gcc7 branch in our Git repo:
<nckx>(Replace /etc/guix/system.scm with your favourite colour bike shed.)
<lfam>isengaara-tomoko: The core-updates branch updates the default GCC to GCC 7
<lfam>isengaara-tomoko: On that branch Guile does build successfully (I used it yesterday)
<isengaara-tomoko>ill try this one, and come again later
<lfam>isengaara-tomoko: The wip-gcc7 branch was deleted on February 10, probably when the changes were merged into core-updates
<bitmapper>i see that libcxx was added
<bitmapper>but i don't see it in the file
<isengaara-tomoko>I had cloned it about one week ago
<lfam>isengaara-tomoko: Okay, sorry for the confusion
<bitmapper>why cant i edit files i open with guix edit
<nckx>bitmapper: Where do you see it was added?
<Elon_Satoshi>bitmapper: because those files are part of the system, you'll have to create a git clone of guix somewhere and then edit the files there
<nckx>bitmapper: Approved, but not added :-)
<nckx>^ maybe rekado forgot
<lfam>Is it worth it to switch from python to python-minimal in native-inputs of packages where it works? It's a small difference in the closure size
<quiliro>nckx: thanks
<quiliro>will try to test it after icecat install...i am dealing with a core2duo here
<quiliro>not much power
<Formbi>does someone here know mcron?
<nckx>quiliro: Same here. Note that it will probably print *a lot*, because Guix doesn't have a tidy notion of human-friendly ‘packages’ anymore at that level.
<efraim>What are you trying to do with mcron?
<Formbi>I wanna make something run when I wake up the computer
<mbakke>lfam: Preferring python-minimal sounds like a good idea.
<efraim>Like @reboot or wake from suspend
<quiliro>nckx: it is rather not printing a lot because it is taking a lot of time to compile rust....weird rust is a dependency for icecat
<Formbi>from what I've read, mcron can detect that the computer was woken up, but I don't know how to use it (if it's even available as an option)
<lfam>mbakke: Okay, I will probably be pushing commits to core-updates later today
<efraim>Mcron doesn't yet have an @reboot, but it will run jobs which were missed while sleeping
<nckx>quiliro: Part of Icecat is written in Rust. Not that weird.
<quiliro>Formbi: mcron uses the same format of cron i recall
<quiliro>nckx: oh"
<nckx>There's probably a way to print the generated derivations without actually building everything, but I don't know it.
<quiliro>nckx: describe?
<quiliro>guix describe
<Formbi>quiliro: does normal cron have an option to run things only when the computer wakes up?
<mbakke>nckx: I use `guix build -n` to print derivations.
<quiliro>i really do not remember much of cron as of now
<quiliro>it is weird... i can telnet port 80 of some servers but epiphany (web browser included with gnome desktop configuration) will not open them sometimes
<quiliro>how can i know which times? is there a log for epiphany?
<bitmapper>i've written a packages and added it to the package path, but it's not showing up in search
<lfam>bitmapper: Is it found by any of the other commands? For example, can you build it?
<lfam>Can you `guix package --show=foo` it?
<bitmapper>i can't build it
<bitmapper>i've got the define module correct
<bitmapper>and the define-public
<lfam>What do you mean by "added it to the package path"?
<lfam>It sounds like something is set up correctly
<bitmapper>i know it works because i installed something else from that directory
<lfam>I mean, "isn't set up correctly"
<lfam>Try adapting this example:
<lfam>In that case, the 'pkgs' directory is at ~/pkgs and GUIX_PACKAGE_PATH is /home/lfam/pkgs
<mbakke>lfam: Should we change the default Python in (guix build-system python) too?
<lfam>mbakke: Hm... maybe? I don't have a sense of whether or not most things will work
<bitmapper>fixed it
<nckx>Turns out Guix gets a bit slow on spinning discs when your store is >1 TB.
<lfam>nckx: Yikes :/
*nckx gcs.
<lfam>That is truly humongous
<nckx>lfam: Thanks.
<nckx>4 years of work, give or take.
<lfam>Just last night I ran gc on a 100 GB store for the first time after deleted years of system generations. I thought that was a big deal but...
<nckx>I guess btrfs dislikes huge directories? So much for btr.
<lfam>btrfs can get really slow in certain situations, although I haven't noticed it in the last several kernels (maybe since 4.4-ish)
<lfam>The dreaded btrfs-transacti
*nckx wonders if /gnu/store/ab/cdef...- would solve this, but mainly why clever fs folks still haven't.
<nckx>lfam: Oh, yes, I remember that...
<mbakke>lfam: I think it might be better to stick to the same Python as the build system to avoid duplication. The comments in python.scm says the -minimal variants are there mostly to break a dependency cycle.
<mbakke>nckx: Woah, 1TB!! I'm jealous.
<lfam>mbakke: Good point
<mbakke>I wonder how much is saved by hard links.
<nckx>mbakke: Crap, I forgot to check before gc'ing.
<nckx>I'd be curious too.
<mbakke>I guess you'll find out soon enough ;)
<mbakke>I find btrfs gets slow pretty fast unless you scrub frequently.
<nckx>I'll just extrapolate whatever's printed at the end (’...currently saves...’). It's bound to be spectacular.
<lfam>I thought that scrubbing was just an integrity check
<nckx>mbakke: It's btrfs + mdraid, both levels scrubbed on the reg.
<nckx>Maybe balance was meant?
*nckx knows it's overkill but wanted raid6 *and* checksums *and* their data...
<nckx>Turns out btrfs RAID6 treats the latter as optional.
<nckx>The space savings are impressive.
<quiliro>build breaks at rust when installing icecat
<kmicu>nckx: 1TB is a huge directory for btrfs‽ xD The issue must be somewhere else.
<nckx>note: currently hard linking saves 5567.20 MiB / guix gc: freed 1357617.60938 MiBs.
<nckx>Implies the original savings could have been ~600 GiB. Or not.
*kmicu keeps ~10TB of data around including a lof of those pesky git kernel repos with plenty of files. No issues at all.
<nckx>kmicu: Hmm. My 8 years of btrs experience have taught me that it's *always* the first suspect.
<lfam>1TB is not a huge directory but the way that Guix works is quite slow on spinning disk with that much data
<nckx>kmicu: Sure, but how many direntries at the top level? I think that's the issue. The 1 TB was for scale.
<lfam>It was really slow even at 100 GB
<nckx>And yes, /var/guix is fully nocow'd.
<kmicu>There are plenty of serious (and happy) NixOS users on btrfs so I doubt that’s the case.
<kmicu>We need benchmarks though for anything constructive. 🤷
<nckx>IMHO btrfs sucking at scale is the null hypothesis, but 🤷 indeed. My store is now 12G and Guix is fast again.
<nckx>(sqlite could also be the culprit, but NixOS uses that too.)
<raghavgururajan>@ncks: Thanks!
<raghavgururajan>1) I installed openvpn, network-manager, network-manager-applet and network-manager-openvpn. But I am not able to import openvpn profile via networkmanager applet. Openvpn option is not appearing in the networkmanager at all. What should I do? 2) Bluetooth is working when I login via root user. But when I login as different user, there is no bluetooth option in networkmanager applet. If I go go bluetooth option in system settings, it says no
<raghavgururajan>bluetooth found.
<Elon_Satoshi>lfam: I could't run guix build qemu-minimal-2.10.2 but I ran guix system --verbosity=5 --no-bootloader init etcetc, and now my computer is frozen but I can see at what point of the qemu-minimal compilation it freezes up. Here's a screenshot I took with my phone:
<lfam>Elon_Satoshi: Thanks. It's hard to debug it without being able to reproduce the issue
<lfam>My recommendation is to install Guix with substitutes and debug it from there, or install Guix on another GNU / Linux distro and debug it from there
<lfam>If I understand correctly, there is a substitute for this package on <> which is one of our two official build farms
<Elon_Satoshi>lfam: Why doesn't have a binary? Maybe they're experiencing the same issue, which might help to reproduce it
<lfam>Elon_Satoshi: Yes, maybe, but the user interface for doesn't make it easy to find out
<lfam>I filed a sort of "UI bug" yesterday: <>
<Elon_Satoshi>Can you ssh into and run `cat $(find /var/log/ | grep qemu-minimal-2.10.2)`?
<lfam>No, I don't have SSH access
<lfam>Or any special access to that farm
<Elon_Satoshi>So everyone else is able to compile qemu-minimal-2.10.2 without problems?
<lfam>I suspect that many people used a substitute
<lfam>This package is a build-time dependency of GRUB, so in practice you don't even need it if you use a GRUB substitute
<Elon_Satoshi>Hmm. Maybe I could give instructions to try to reproduce the issue
<lfam>Elon_Satoshi: I did read this log-file of the failed build of the *current* qemu-minimal-2.10 package:
<lfam>I don't know whether or not that is the same package you are trying to build. Months have passed since 0.16 was released
<lfam>Looks like the interesting stuff is not displayed there, however
<Elon_Satoshi>lfam: I have encountered this problem several times after running guix pull, then logging out and running guix --version to confirm that I'm running the latest guix commit
<bavier>I just built qemu-minimal-2.10 locally and had no issues
<bavier>didn't have swap enabled, noticed no hiccups
<lfam>Elon_Satoshi: I don't recommend using `guix pull` before installing a new system. It voids the warranty, so to speak
<lfam>We test installation of the system using the installation image as it is provided. Using `guix pull` voids that testing
<quiliro>what can i do to avoid this build error?
<Elon_Satoshi>bavier: How did you build qemu-minimal-2.10 locally? `guix build qemu-minimal-2.10.2` or even `guix build qemu-minimal@2.10.2` doesn't work for me
<bavier>Elon_Satoshi: guix build --no-substitutes -e '(@@ (gnu packages virtualization) qemu-minimal-2.10)'
<mikadozero>raghavgururajan: why not use the command line instead of the gui to connect with openvpn? sudo openvpn /path/to/openvpn-config-file.ovpn
<mikadozero>raghavgururajan: you can probably get the openvpn .ovpn config file from your vpn provider.
<quiliro>why do i get this rust build error when installing icecat?
<quiliro> does not open on epiphany...does this happen to you people too?
<Elon_Satoshi><lfam> Elon_Satoshi: I don't recommend using `guix pull` before installing a new system. It voids the warranty, so to speak
<quiliro> enters now
<Elon_Satoshi>Alright, I'll try running guix system init without running guix pull
<Elon_Satoshi>bavier: Can you try locally building qemu-minimal-2.10.2 on a GuixSD 0.16.0 livecd?
<CornBurglar>I'm trying to use the guile-cairo library that I installed with guix, and (use-module (cairo)) is not finding the module. Is there a path I need to set or something for it to be recognized globally?
<Elon_Satoshi>With a hard drive mounted to /mnt and the cow-store service started
<Elon_Satoshi>In fact I think I'm gonna do that
<CornBurglar>To be clear I am on GuixSD
<Elon_Satoshi>bavier: I'm getting an error: "guix build: error: (@ package not found for version (gnu packages virtualization) qemu-minimal-2.10)"
<bavier>Elon_Satoshi: the quotes are needed
<Elon_Satoshi>the quotes are there
<Elon_Satoshi>the command was `guix build '(@@ (gnu packages virtualization) qemu minimal-2.10' | tee /mnt/guix/qemu-minimal-2.10-build-fail.log`
<Elon_Satoshi>actually the command was `guix build '(@@ (gnu packages virtualization) qemu minimal-2.10)' | tee /mnt/guix/qemu-minimal-2.10-build-fail.log`
<bavier>Elon_Satoshi: you're missing the -e flag
<Elon_Satoshi>oh okay
*bavier realizes they missed testing a few packages in the openmpi upgrade: scalapack, petsc, and dealii broken...
<Elon_Satoshi>just ran `guix build '(@@ (gnu packages virtualization) qemu minimal-2.10.2)' | tee /mnt/guix/qemu-minimal-2.10-build-fail.log` and got "in procedure module-lookup: Unbound variable: qemu-minimal-2.10.2"
<Elon_Satoshi>huh, so 2.10.2 doesn't exist but 2.10 does?
<bavier>Elon_Satoshi: which the `guix build -e ...` syntax, you're asking to build the package a scheme variable evaluates to; in this case, the scheme variable name includes a version component
<gururajanraghav>Can users' files and data get erased after "system reconfigure"??
<quiliro>why are some websites not available on epiphany but with telnet website 80 they are available?
<Elon_Satoshi>I hope 2.10 is the same thing as 2.10.2
<quiliro>on GSD
***gururajanraghav is now known as raghavgururajan
<Elon_Satoshi>quiliro: an example of such a website?
<Elon_Satoshi>also try using curl on a website instead of telnet, that way you don't have to go through manually typing the whole hypertext transfer protocol
<Elon_Satoshi>`curl` should display all the HTML on that website
<Elon_Satoshi>bavier: I am now going to run `guix build -e '(@@ (gnu packages virtualization) qemu-minimal-2.10)' &>/mnt/gu/qemu-minimal-2.10-build-fail.log` on guix version 0.16.0-1.c845323
<Elon_Satoshi>it's downloading a substitute from
<Elon_Satoshi>why didn't it do that when I was running guix system init?
<Elon_Satoshi>using substitutes
<Elon_Satoshi>And now I'm trying to build it locally and it's downloading the substitute for some reason
<Elon_Satoshi>did successfully build qemu-minimal-2.10.2 just now?
<bavier>Elon_Satoshi: it's possible that it took some time for to "bake" the substitute
<bavier>for the particular substitute you're requesting
<Elon_Satoshi>bavier: the command finished without compiling anything
*Elon_Satoshi now runs `guix build --no-substitutes -e '(@@ (gnu packages virtualization) qemu-minimal-2.10)' &>/mnt/guix/qemu-minimal-2.10-build-fail.log`
<quiliro>curl http://freedombone.local/admin
<quiliro>curl: (6) Could not resolve host: freedombone.local
<quiliro>but telnet freedombone.local 80
<quiliro>will work
<Elon_Satoshi>quiliro: weird
<Elon_Satoshi>I tried telnet freedombone.local 80 over here and got "telnet: could not resolve freedombone.local/80: Name or service not known"
<Elon_Satoshi>but, is that a local address on your computer?
<Elon_Satoshi>quiliro: do you have a public website where you experience this issue?
<quiliro>Elon_Satoshi: it is a local site...i am trying to find a public one
<quiliro>the same server is
<quiliro>but i cannot access the same things from that domain
<quiliro>i can access that domain locally
<quiliro>perhaps something to do with problem viewing html5 videos on youtube
<quiliro>on epiphany
<rekado> says “403 forbidden”
<ZombieChicken>Hello. I'm trying to add a mapped-device to my system config and i"m unsure of the format needed. Do I need to use a cons* for a single entry in the mapped-devices ...) portion?
<rekado>ZombieChicken: “cons*” is used when you want to add one or more items onto a list.
<rekado>it’s like “cons” which only works for a single item and a list.
<ZombieChicken>yeah, I'm aware of that, but wasn't sure if that was the proper option since it feels rather awkward
<rekado>what feels awkward…? Could you show an example?
<ZombieChicken>using (cons* ...) for a single entry in a list
<ZombieChicken>I think most of my problems atm are related to guile and not guix atm
<rekado>do you mean (cons* (mapped-device …)) ?
<rekado>because that won’t work
<rekado>cons* needs at least two arguments.
<rekado>if you want to define a list use “list”.
<ZombieChicken>quote wouldn't work?
<rekado>'(anything) is the same as (list 'anything)
<rekado>if you want a non-symbolic value in the list you shouldn’t use quote.
<ZombieChicken>yeah. Seems to be some differences between how racket does things and how guile does that is messing with my head a bit
<ZombieChicken>but I've got this thing working I think
<ZombieChicken>will find out upon reboot
<rekado>I don’t think they are different in that regard.
<raghavgururajan>Can anyone please help with my problem? Thanks!
<ZombieChicken>so apparently no automatic encrypted root detection
<quiliro>how about
<rekado>quiliro: tells me that something’s wrong with the certificate
<quiliro>i get 403 with too
<rekado>raghavgururajan: consider writing to with sufficient detail
<quiliro>it is a freedombone box
<rekado>quiliro: 403 is a server response. This is nothing to do with your browser.
<raghavgururajan>@rekado: Oh, I didn't know about that. Thanks a lot! I'll do it now.
<quiliro>true but on i get:
<quiliro>Información técnica
<quiliro>► visible
<quiliro>P Atrás
<quiliro>that weird text
<quiliro>perhaps missing images
<quiliro>it says nothing about certificates... how can i see that?
<quiliro>it is i can enter freedombone.local/admin
<quiliro>why is this? because i tryed, which is the same machine?
<quiliro>this happens to me every day
<quiliro>then access is ok
<apteryx>hello! any recommendation for a home wifi AP/router?
<apteryx>So far most capable option seems to be this:
<bavier>apteryx: I ran one of the thinkpenguin routers for a while, I've also got a buffalo wzr-hp-g300NH that I've been meaning to put librecmc on
<quiliro>i cannot install guix system distribution for an end user because they cannot see videos
<rekado>I also got that buffalo (could be a single letter difference in the type, but looks close).
<rekado>quiliro: with icecat video playback does work.
<rekado>for epiphany there’s an open bug report relating to reliable video playback.
<apteryx>bavier: is installing libreCMC as easy as installing dd-wrt/openwrt (select a firmware file to upload from the web admin/mgmt page of the device?), or do I need some JTAG programmer or the likes?
<bavier>apteryx: I think you can do the firmware upload thing
<mikadozero>raghavgururajan: For openvpn have you tried the command line instead of the gui? sudo openvpn /path/to/openvpn-config-file.opvn You should be able to get the .opvn file from your vpn provider.
<apteryx>the option from Minifree is interesting (only 35 EUR), but the shipping to the other side of the Atlantic would cost me 45 EUR (and this is before duties come in)
<pkill9>what's the option from minifree?
<ng0>is that libreCMC still being developed?
<ng0>last time I switched from it, it was seriously lacking behind their supposed upstream
<mikadozero>I have guile-wm questions.
<mikadozero>1. Why does the installation put a .guile-wn in my user and /root directories I did not add any window managers in my config.scm?
<ng0>they are part of the profile skeleton
<gururajanraghav>@mikadozero: No, I haven't. But I'll try.
<mikadozero>ng0: Is there documentation on the profile skeleton?
<ng0>the source which generates it. one mom
<bavier>ng0: librecmc's latest release dates 2019-01-12, so, yeah, it's still being worked on :)
<ng0>aha, "again". it used to be silent for a long time
<mikadozero>When I run which guile-wn there in no program installed. Is guile-wn run by having guile execute the .guile-wn?
<ng0>no, it should be picked up by a combination of having guile-wm and its dependency in the system definition
<ng0>then it can simply started via the login manager
<ng0>but I won't recommend guile-wm, unless you want to fix it yourself
<ng0>upstream is basically dead silent, and the WM dies after a while into the session depending on the key combinations
<ng0>at least that's my experience
<ng0>iirc there should be documentation about the skeleton when you search for "skeleton" in the info documentation for g uix
<ng0>gnu/build/activation has the code iirc
<ng0>or gnu/system/shadow
<ng0>my mind's elsewhere right now
<mikadozero>ng0: Thanks
<ng0>not activation
<ng0>i mean the idea of guile-wm is interesting, but it doesn't help when upstream seems to have silently moved on to other hobbies. adopting it would be nice, but someone's got to do it
<ng0>except for configuration being done in guile, there's not much gain over cwm for myself. and cwm works.
<ng0>but last time I checked out guile-wm was 6 months ago or so, maybe the situation changed
<ng0>what I was refering to, a while back:
<mikadozero>ng0: I see there has not been any response to your issue 11 in more than a year.
<roptat>I'm having some troubles parsing that sentence in the manual : "This is achieved by passing the @code{#:name-services} argument to @code{nscd-service} the list of packages providing the needed name services"
<ng0>and no commits since march 2017. no big deal, like people move on but it would be nice to at least announce it. I know I'm doing this myself too, but you need the time, energy, etc for free work
<mikadozero>ng0: Yes it is nice when an announcement is made.
<gururajanraghav>How to add hostname along with domain name? For example, in other distros, there will be field to mention in the format of "hostname.domain.tld".
***gururajanraghav is now known as raghavgururajan
<lfam>gururajanraghav: You can set the hostname in the system configuration (config.scm)
<nckx>raghavgururajan: Hm? (host-name "").
<raghavgururajan>@lfam: but the field the config is just hostname right? not hostname.domain.tld?
<nckx>raghavgururajan: There is no difference. Hostnames can be FQDNs. Probably should be.
<nckx>A small minority of software will choke on unqualified hostnames.
<raghavgururajan>Kinda confused. I'll read the contents on the link and get back here for better clarity. XD
<raghavgururajan>@nckx thanks
<nckx>From my config: (host-name "") ; it's really that easy ;-)
<nckx>Interestingly (?), ‘hostname’ returns it, but ‘hostname -s’ returns ’localhost’. Not sure if that's expected.
*nckx hashtag shruggaroo
<janneke>vagrantc: i have resurrected the debian packages for mes, mescc-tools and nyacc -- it would be great if you could have another look at them
<Elon_Satoshi>now guix is downloading python 3.7.0 twice ._.
<lfam>Elon_Satoshi: There are both python and python-minimal packages. There are also wrapper packages for each of those
<Elon_Satoshi>nope they're both python-3.7.0 and they're both 20.6MB