IRC channel logs


back to list of logs

<bt`>nckx: it seems to give the same error after the pull. A while back I pulled to a commit, I suspect that's not a problem but thought I'd mention it just in case. I haven't given my system a chance to cool off but it already seems much much better thanks guys.
<rekado_>pkill9: likely out of memory
*rekado_ –> zzzZ
<pkill9>ah that's annoying
<pkill9>rekado_: it fails on the berlin build farm
<nckx>bt`: That shouldn't™ affect your current guix generation. So you've done the equivalent of ‘guix pull && guix build thermald’? Then I'm very interested in your build output.
<pkill9>this is the failure log incase anyone is interested reading this
<bt`>I can get you the log nckx
<nckx>bt`: You can use cpufreq-set -u … to underclock your CPU even when fully loaded, until you get thermald working.
<bt`>It seems to be cooling off fine now, I suspect changing the governor and killing the run away slim was enough. I need to upload the log I'll be right back.
<nckx>(Thermald will have the same negative effect on long-term performance, but will allow for high-frequency bursts that will probably significantly improve UX.)
<nckx>(There's always vacuumcleanerd and reapplythermalpasted, too…)
<bt`>I badly need to re-automate uploading files to my server.
<nckx>scp not good enough for ya'? (Thanks.)
<nckx>bt`: What commit does ‘guix describe’ print?
<bt`>guix describe: error: failed to determine origin
<bt`>that seems fun
<bt`>hint: Perhaps this `guix' command was not obtained with `guix pull'? Its version string is 1.0.0.
<nckx>We might be on to the bug here ;-)
<nckx>bt`: ‘command -v guix’?
<nckx>bt`: Bingo.
<nckx>You have, accidentally or on purpose, added guix to your system-packages. (??)
<nckx>Is that possible?
<bt`>How would I have gone about doing that, I don't think I added that to my config ever.
<nckx>I'm not quite sure how that would happen either.
<bt`>just (cons* nss-cert emacs emacs-exwm %base-packages)
<nckx>Welp. Does ‘which -a guix’ print any more guixen?
<nckx>(It should.)
<bt`>that's it
<nckx> bt`: Can you run /home/$USER/.config/guix/current/bin/guix ?
<bt`>I'm running this all as root is that incorrect?
<bt`>yes yes it is
<nckx>bt`: It's not great. May I ask why (and how; sudo)?
<nckx>Unless you're actually logged in as root, you should just run guix as yourself.
<bt`>I just sudo su, because I'm used to machines without sudo :p anyway the response to your questions is now different one moment
<bt`>bt@de ~$ command -v guix -> /home/bt/.config/guix/current/bin/guix
<nckx>OK, that's the correct (pulled) guix.
<bt`>bt@de ~$ which -a guix -> /home/bt/.config/guix/current/bin/guix /run/current-system/profile/bin/guix
<nckx>Unless you know what you're doing™, you should just run ‘guix’ as yourself.
<bt`>same error when running package build thermald, the reason I was running as root was to reconfigure my system so I just didn't sign out I guess.
<bt`>guix build thermald*
<nckx>The only exception is ‘sudo guix system reconfigure …’ (because it has to install a boot loader at the very end), but even that uses your user's guix; it's not equivalent to ’sudo su - … guix’. I don't know if there's a way to use guix with ‘su’ like that.
<bt`>could that have caused this issue?
<nckx>bt`: But you can ‘guix describe’ now, correct?
<bt`>Generation 2 May 17 2019 23:22:41 (current)
<bt`> guix 08eafef
<bt`> repository URL:
<bt`> branch: master
<bt`> commit: 08eafef8650922681bad10d9f5a3f92ff91616b6
<nckx>It would lead to exactly this problem… *but* then ‘guix build …’ should have worked as bt.
<nckx>And you said it doesn't so that's strange.
<nckx>Let me build thermald on that exact commit just to make sure that it actually works for me.
<bt`>could the build log be different when run as the user? would that be useful?
<nckx>I can't say. You say you ran (as bt) ‘guix build thermald’?
<nckx>If that failed and there's a log it would be nice to see.
<nckx>Even if it's the same error the hashes might be different.
<nckx>And I'll at least know we're building the exact same thing.
<nckx>bt`: Something is still wrong. You're building /gnu/store/dq6y934sls06g1pq23bxm826sv15fn7k-thermald-1.8 (the value of --prefix in the log).
<nckx>When I ask guix@your-git-commit to build it:
<nckx>The store hashes should be identical. Guix doesn't make mistakes (only its authors do).
<bt`>(and users)
<nckx>(and CPU manufacturers)
<nckx>To top it off, I'm afraid I really have to go to bed.
<nckx>I think you should report this to
<bt`>well that's fine nckx, I really appreciate all you've done. sleep well.
<bt`>I can do that nckx
<nckx>Good luck, and good night.
<bt`>thanks see you around.
<pkill9>does the opam importer work? it fails to download meta-data for me
<Marlin1113>where are the kernel modules located on guix?
<noobly>is anyone working on hurd
<noobly>oh but I came here to ask how emacs-guix works for emacs package management, does it replace the use-package interface and melpa or what?
<calher>noobly, it should replace the Emacs package manager, yes.
<Marlin[m]>Hey guys
<Marlin[m]>So, i'm trying to install again
<Marlin[m]>but i get no video at all
<Marlin[m]>So the GUI installer doesn't work
<Marlin[m]>I though starting X could help, as that was probably a Framebuffer issue
<Marlin[m]>But it seems like X doesn't recognize my screen
<calher>Has anyone started a recipe for Telegram?
<archetyp>Currently I use Telegram with Pidgin, I would be happy about the original client.
<calher>Why is Guix so slow to start the GDM?
<calher>I always think there's no GUI, it takes so long.
<emacsite>I've run into a problem.
<emacsite>My GNOME session freezes now, whenever I do Logo+Space to switch from UK QWERTY to Hebrew keyboard layouts.
<emacsite>And it takes forever to boot my system.
<emacsite>hi pagurus`
<brendyyn>i notice that after i type strace there is no tab completion on guix system, but there is on Arch
<calher>Help. GNOME's menu and Alt+F2 aren't responding.
<calher>I've rebooted. I've restarted xorg-server. None of these have worked.
<calher>And why is irssi beeping at me when nothing new has printed to the screen? :(
<xelxebar>Anyone here tried out one of the pre-built qcow2? I'm having trouble logging in. Apparently, root has a password set by default and I can't find what it is.
<calher>I give up. Going back to Trisquel. I can't even get X to stay open.
<calher>and it takes forever to boot and launch x
<brendyyn>xelxebar: i think there is no root password and you have to set iit
<buenouanq>it's so funny and strange, GuixSD is the first distro (besides pre systemdebian stable) that I haven't had any real issues with. Everything just works how it should and how I expect.
<buenouanq>When people can't get things working on it, I'm absolutely boggled.
<buenouanq>I want to help, but I never encounter the same issues and have no idea where to point people ( ._.)
<buenouanq>I think there might be some sort of systemd/arch/gentoo stockholm syndrome where people think things should be way more complicated and involved than Guix makes them.
<xelxebar>brendyyn: Hrm. That's what I thought, too, but I can't seem to log in. It's an old image. Perhaps I set the pwd way back when. Thanks for confirming
<calher>buenouanq: I was willing to deal with the slowness of the system booting up, were it not for other issues. GNOME no longer gives me a menu when I pres Logo or click on Activities. GNOME freezes on me when I switch keyboard layouts. Reboots and restarting X do not change these results.
<calher>buenouanq: I am using an X200T with Libreboot, and I chose the disk encryption option in the installer.
<brendyyn>calher: did you file a bug report?
<calher>brendyyn: No, but I'll try my best to describe the problem and what I did.
<brendyyn>im sure you can get more assistancee if you explain everything on the mailing list and are patient.
<calher>What is the email address?
<calher>Thanks. Should I keep this system on another disk for testing?
<brendyyn>i dunno. maybe there are logs that you can extract that tell you whats going wrong
<calher>So yes.
<calher>I guess I'll dump this system to a disk.
<calher>Otherwise, I'm very impressed with Guix so far. It has what I need: easy installer, easy disk encryption, qTox, HexChat, Mumble, qBitTorrent or Transmission, a reasonably up-to-date Mozilla browser, LibreOffice, youtube-dl, Tor, KeePassXC, Hebrew input, Russian input, Gajim...
<calher>Missing (but not a big deal): Telegram Desktop, Thunderbird or Evolution, X200T touch screen support, Transmission GTK, GNOME Software integration,
<buenouanq>transmission gtk is there
<buenouanq>you install it with transmission:gui
<buenouanq>didn't mozilla kill thunderbird? I've moved to claws-mail and it's pretty great.
<brendyyn>they did not kill it, they just stepped away from it officially
<calher>Mozilla emancipated Thunderbird.
<brendyyn>it still exists just like any other community based free software project i guess
<calher>No more 'dead' than Claws.
<calher>I wonder... can Guix tarball packs run on Windows Subsystem for Linux?
<brendyyn>one way to find out
<calher>I don't own a WC.
<calher>I don't have a WC.
<calher>I have a stdout house.
<brendyyn>cat dog > house
<g_bor[m]>Hello guix!
<g_bor[m]>I am having a problem with one of my systems. I get an error on login, that cannot make/remove session, and I get dropped back to the login prompt.
<g_bor[m]>Any idea how to fix that?
<Blackbeard[m]>g_bor: is this on a tty or desktop environment
<Blackbeard[m]>g_bor: did you change something?
<Blackbeard[m]>maybe your .bash_profile
<g_bor[m]>it's on guix system, after a fresh pull. I noticed this sporadically eariler, but on this new version it seems that dbus-system consistently fails to start on boot. I tried several earlier generations, and they had the same problem, but from a generation for March I managed to log in after failing the same way several times.
<g_bor[m]>I did not change any of my dotfiles recently.
<Blackbeard[m]>g_bor: what happens if you try to login as root
<g_bor[m]>the same thing
<g_bor[m]>I do not get to the desktop.
<g_bor[m]>I believe it's a service startup issue.
<Blackbeard[m]>g_bor: do you get errors on 'guix system reconfigure'
<g_bor[m]>I believe I don't, but I will do a pull and reconfigure to check
*g_bor[m] afk, will report back later
<roptat>hi guix!
<Formbi>could manifests be modified so one could _add_ the packages listed in a manifest instead of installing only these packages?
<civodul>everything is possible but the idea of manifest is to declare what you want
<civodul>so we'd have to discuss this use case, it's not obvious to me that it's one we'd want to support
<jonsger>morning guix
<Formbi>I get it, the thing I'm talking about could be something like --add-from-manifest
<brendyyn> so many flags :/
<efraim>--I-like-long-flags-and-I-cannot-lie --lets-try-experimental-features --yes-I-am-really-sure-this-is-a-flag --simon-says-sudo
<pkill9>im finding guix is hanging when trying to access this inferior package by running `guix build -f <file-from-the-debian-paste>` (does the same when it's in a manifest):
<pkill9>anyone have any idea?
<pkill9>woo i tried another commit and it worked
<pkill9>i guess handling inferiors just needs to be improved, some commits seem to hang
<g_bor[m]>hello guix!
<g_bor[m]>I remember that I have asked this before, but I just can't remember how can I find the packages inherited from a package.
<archetyp>calher: The GDM probably has problems with some external monitors on ThinkPads.
<pkill9>berlin gateway timeouts :<
<g_bor[m]>hello guix!
<g_bor[m]>On one of my systems dbus-system fails to start on boot. This is a very painful point, as you simply cannot log in from then on, since logind cannot connect to dbus, therfore the PAM module returns an authentication failure. It also fails to bring on networking from then on, since it also depends on dbus-system. Any idea how to debug this?
<lurker>does GuixSD support ZFS and can it be installed on ZFS?
<nly>qemu-system-x86_64: failed to initialize KVM: Permission denied
<g_bor[m]>lurker: ZFS has a license incompatibility with the linux kernel. Until that is resolved Guix System will not provide it out of the box. Nothing stops you from creating a setup where ZFS works, but it would take some work.
<g_bor[m]>nly: what way are you starting the vm? Would prepending sudo to the command help?
<lurker>g_bor[m]: what work needed?
<g_bor[m]>lurker: most notably a custom kernel.
<lurker>g_bor[m]: are there ready to use configurations of custom kernel for guix?
<g_bor[m]>The problem is that you can't distribute the zfs kernel module. See
<lurker>i think something like AUR with option to use proprietary packages would be nice for Guix ecosystem
<g_bor[m]>lurker: I don't believe so, at least on on the zfs front.
<nly>g_bor /gnu/store/
<nly>sudo helps
<Marlin[m]>Guix won't even show the installer for me
<Marlin[m]>i'm gonna cry
<g_bor[m]>lurker: Now we have channels support, and a bunch of us has channels with proprietary, or otherwise non upstreamable packages, but Guix System is an FSDG disto, which means that proprietary software is not an option for upstream.
<g_bor[m]>nly: if you also have root access can you try it with sudo? It does work me that way here. I don't know if we have any special groups that would allow running it as a simple user.... Maybe.
<g_bor[m]>Marlin: how could we help? Where do you get stuck?
<Marlin[m]>it's an issue with my gpu still
<Marlin[m]>It just doesn't show the installer
<Marlin[m]>I tried starting X from the usb drive, it doesn't even recognize my screen
<g_bor[m]>lurker: but recently a blog post was written on the topic of creating custom kernels.
<lurker>g_bor[m]: i'm okay with upstream being fully floss and not proprietary. centralized channel with user custom definitions is imo an important feature. where to find existing channels?
<g_bor[m]>Marlin: do you get to grub? Can you get past that? If yes, there should be root consoles on tty4-6 at least.
<Marlin[m]>i did change to root consoles
<lurker>g_bor[m]: could you post link to blog post?
<Marlin[m]>Manual installs aren't for me tho
<Marlin[m]>Last time i got it wrong
<g_bor[m]>lurker: I am at it. If you are interested in a channel that provides 'more', then try searching for it :) Please take this as a hint...
<g_bor[m]>Currently there is no central place where channels are collected.
<janas>It would be nice to have a channel listing somewhere, hopefully searchable by package name
<g_bor[m]>lurker: have a look at this, this a blog post draft, and is not published yet.
<lurker>g_bor[m]: thanks!
<g_bor[m]>lurker: yw!
<pkill9>is the cuirass REST api documented anywhere?
<g_bor[m]>pkill9: partially. It is mostly the same as the hydra api, with some small extensions for the web interface. Hyda api is documented I believe.
<g_bor[m]>What are you trying to do with it? It is currently really slow, some queries just consistently gateway timeout.
<pkill9>g_bor[m]: i want to find the latest guix revision that a given package was successfully built with
<pkill9>g_bor[m]: do you kjnow where the manual for hydra is hosted? i found the source code for the manual
<pkill9>g_bor[m]: basically i want to get these search results in a script-friendly way
<pkill9>then get the git commit of the guix revision it was built with
<pkill9>(i can't see that in those results though, so i may have to guess based on the version of the package and/or date the build was made)
<nckx>lurker: Note that any such listing would still not include non-free channels.
<nly>for dvorak layout it's (keyboard-layout "dvorak")?
<nckx>nly: "us" "dvorak", I think.
<lurker>nckx: why not include non-free? why not to have listing of *all* channels?
<nckx>lurker: Because we're a GNU project and don't promote proprietary software.
<pkill9>hmm, running `curl -i -H 'Accept: application/json'` returns HTML
<lurker>nckx: can alternative listing exist, that won't be a part of GNU project?
<nckx>lurker: Sure, like those channels themselves already exist, anyone can create a list (as long as they make it clear that it's not official, for example by not using the name ‘Guix’ in a misleading way). But the Guix project itself won't link to them or recommend them, and in this channel friendly people like me will ask you not to do so ;-)
<lurker>nckx: that means in this irc channel non-free listings are forbidden to be shared?
<nly>what's the snippet to use slim instead of gdm login manager?
<nckx>lurker: Yes, because we consider proprietary software harmful.
<nckx>It can be tempting to answer ‘halp my wifi doesn't work’ with ‘just install this [blob]’, with nothing but good intentions, but that's not considered helping here ;-)
<g_bor[m]>nckx: I was thinking about this ovmf stuff, and a possible working compromise would be to unpack the source of the system openssl package in place of the submodule. That would work, but I don't know if it's the right thing to do. Wdyt?
<nly>it's incredible how hard it is to use guix system configuration
<nckx>g_bor[m]: Considering how Guix manages dependencies (‘well’), I don't see how it could hurt. It's just another input. If OpenSSL applies a security fix upstream, our OVMF will always be rebuilt, unless I'm missing something…?
<nckx>g_bor[m]: We do the same thing with googletest (not the best comparison but the only one that comes to mind) and it works.
<nckx>g_bor[m]: Anyway, I think you're on to the best fix here.
<g_bor[m]>ok, thanks :)
<g_bor[m]>nckx: do you have any idea how to list the package inheriting from a package?
<g_bor[m]>it seems that guix refresh -l misses those.
<nckx>g_bor[m]: No! I've been wondering that myself.
<civodul>the search feature in Cuirass is neat, kudos rekado!
<nckx>I thought about it a year or so ago and my idea was to have (inherit …) set some kind of explicit ‘inherited’ flag.
<g_bor[m]>civodul: It gateway timeouts here :(
<nckx>Now I'm not so sure.
<bitmapper>hey, i'm trying to write a guix package for a precompiled binary (it is open source, and i feel comfortable using the binary, i just don't have the compute power to compile it), what would be the best way to go about patching the paths
<g_bor[m]>nckx: Yes, the problem here seems that inheritance information is not propagated in this case... I believe in the worst case it should be possible to traverse all packages seeing the definition, looking for inherit. This is not a query that end users should be very interested in... so low performance might be tolerated.
<cbaines>bitmapper, there's a program called patchelf that comes in useful here
*nckx is mucking about at the REPL to see if their understanding has improved.
<bitmapper>yeah i know, but what would be the best way to call it
<bitmapper>i'm coming from nixos so this isn't completely foreign to me
<bitmapper> the whole metapackages thing is a bit confusing though
<cbaines>bitmapper, I haven't used it recently, but I did a package definition wrapping the phantomjs binary a long while ago
<cbaines>bitmapper, as for metapackages, what do you mean?
<bitmapper>i've noticed many packages in one file, which is a bit strange coming from nixos
<bitmapper>it seems a bit cluttered
<pkill9>those files are modules
<pkill9>they're basically categories
<pkill9>categories of packages
<pkill9>well, the modules are guile modules, but they are used as a way to put packages into categories
<pkill9>i dunno what the rationale behind having a bunch of packages in a single module is, maybe its just a leftover convention from the early days of guix
<cbaines>sometimes I find myself wanting more categories, but actually, that can just make it harder to work out where to put packages/services...
<bitmapper>imo i'd make categories just folders of individual package files
<cbaines>It has been considered I think, but you'd either need to make each file for each package it's own module, or manage loading them in some other way
<bitmapper>well on nixos there
<g_bor[m]>I believe it is for performance reasons. At least that was mentioned when I was asking the same question eariler.
<bitmapper>ahh sorry
<bitmapper>there's a default.nix file with a list of names and package file paths
<pkill9>bitmapper: if you want to be lazy, you can just use patchelf to change the interpreter path, and then run the binary with LD_LIBRARY_PATH to find all the shared library files
<pkill9>if you want to avoid having to write a package definition for now
<g_bor[m]>anyone else had problems with dbus-system not starting on boot, or is this affecting only me?
<bitmapper>i'd prefer a package
<nckx>g_bor[m]: No…
*g_bor[m] afk
<bitmapper>my dog just knocked over my laptop and it landed upside down
<bitmapper>now i need to backup my stuff
<nckx>bitmapper: Ouch. I've been there. Now i back up my stuff.
<rubic88>Are most guix users running icecat as their primary browser? I'm a chromium user, but couldn't get ungoogled-chromium working on my foreign distro.
<bitmapper>i don't use a web browser
<bitmapper>i just log straight into emacs
<nckx>rubic88: I reckon'. (If by ‘primary’ you mean ‘the big GUI one that some sites force you to use’; many Guixers prefer emacs etc.)
<nly>chromium or icecat, others too but they may be too experimental
<bitmapper>my research paper is fine
<rubic88>I use spacemacs, but develop web apps so need at least chromium/firefox equivalent.
<Marlin[m]>rubic88 did you try the ungoogled chromium appimage?
<bitmapper>i don't think appimages work on guix
<rubic88>Marlin[m] I tried installing it. First time it froze my system.
<Marlin[m]>What is the issue with appimages on guix?
<bitmapper>probably the same one on nix
<bitmapper>i don
<bitmapper>* i don
<bitmapper>i need to get used to this keyboard
<nly>nckx i've also developed a habit to rebase often, it's neat!
<bitmapper>i used to use spacemacs but the slow speed bugged me
<rubic88>Just installed icecat on a foreign distro. Everything has weird unicode chars. Is there a secondary package I need to install?
<rubic88>bitmapper spacemacs speed is definitely an issue
<bitmapper>i just wrote my own emacs config instead
<bitmapper>it's been nice
<pkill9>it's possible to get appimages working on guix, i made a service for it :3
<rubic88>guile works nice out of the box with spacemacs, I'm just now learning scheme.
<pkill9>github has a package manager now
<pkill9>or i misunderstand what it is
<bitmapper>it's a repo for npm and yarn
<bitmapper>and docker, maven, nuget + rubygems i guess
<nckx>I can't log into CUPS (:631) on Guix System. I'm a member of lp.
<nckx>‘Authentication failed for user "nckx"’.
<kmicu>rubic88: iirc IceCat needs FontConfig from Guix too.
<rubic88>kmicu: Thanks. Should I install cairo, too?
<kmicu>I don’t think so, those ‘weird unicode chars’ are � (REPLACEMENT CHARACTER) it only means IceCat cannot find fonts. Rendering is ok cuz we see replacement characters.
<nly>error in finalization thread: Success
<nly>After that it hangs
<rubic88>kmicu: Nope, no joy even with fontconfig installed.
<kmicu>rubic88: could you install a font from Guix too and restart IceCat?
<rubic88>kmicu: Sure. Recommendation?
<rubic88>libxfont, maybe?
<kmicu>font-gnu-unifont, font-liberation.
<another-user>i'm trying to build my own pakage but getting following error:
<another-user>no code for module (.cache guix pull pjmkglp4t7znuugeurpurzikxq3tnlaywmisyr27shj7apsnalwq build-aux build-self)
<another-user>how can i fix it?
<nly>Can you paste your PKG def?
<rubic88>kmicu: No luck with font-gnu-unifont, font-liberation, or font-google-roboto. Running from terminal: Pango-WARNING ... failed to create cairo scaled font
<kmicu>rubic88: if there are still no proper glyphs in IceCat then you could also try installing pango (text layout (shaping) and rendering library) from guix.
<kmicu>rubic88: oh, it even warns about Pango. Great.
<another-user>my builder is incorrect but i'm just trying to get at least something to to work with, it doesn't seem that error is related to builder
<rubic88>kmicu: Same problem with pango installed.
<nly>Builder should return #t I think
<bitmapper>you might have to install the fonts using the host package manager
<kmicu> rubic88: besides running ‘fc-cache -r’ or installing cairo I’m out of ideas 🤷
<another-user>nly: i tried but it didn't help, also error message says something about display-error and missing code for module
<nckx>another-user: It says warning:, not error:. The error is what follows.
<rubic88>kmicu: Nope, sorry. Thanks for your suggestions.
<nckx>Not that I can figure out what's causing it.
<another-user>nckx: yeah, error is really obscure, why does something call "display-error" with incorrect number of args...
<kmicu>rubic88: it looks like you need to relognioo.
*nckx nods.
<kmicu>rubic88: (or source ~/.profile from a shell and start IceCat from it.)
<rubic88>kmicu: Works, thanks!
<kmicu>Correction: ~/.guix-profile/etc/profile*
<kmicu>rubic88: glad it works. And sorry for inconvenience but guix packages interacting with host’s runtime are always tricky.
<kmicu>(Ideally eveyrthing should be isolated and Just Work™ after executing ‘guix install icecat’ on a not-Guix-System distro.)
<another-user>i'm trying to build using "guix build -K -L /home/user elasticsearch" command, maybe this is wrong?
<rubic88>kmicu: Understood. I'm not complaining, esp. because I'm working on a foreign distro. Assuming that running GuixSD (soon, I hope) that everything pretty much runs upon installation, correct?
<another-user>elasticsearch.scm is located in /home/user
<nckx>another-user: Guile error reporting is very meh. Things also cascade quickly into seemingly unrelated messages, so it's still worth fixing other issues with your package. Like where bash is supposed to come from.
<nckx>Your expression is also malformed, in a way that you can't really call any errors unrelated: I assume ‘let’ is supposed to be part of ‘begin’, but it isn't, so there's this weird dangly thing in your builder that can easily make things blow up later.
<nckx>(Again, in an ideal world this would be reported, but, well, look out of the window.)
<nckx>If I fix that (by moving a ‘)’ from ‘(use-modules (guix build utils)))’ to the close of ‘arguments’, I get a sane error message \o/
<nckx>Unbound variable: output
<another-user>nckx: oh, you're right, thank you!
<kmicu>rubic88: yep correct. Cuz in Guix System we have everything under our control. Guix-only should work too but some packages, like those sharing graphic stack with host, are more difficult/expensive to fix.
<nckx>Which also makes sense because ‘((source (assoc-ref %build-inputs "source")))’ has one ‘)’ too many. And so on…
<nckx>another-user: Do you use emacs? C-M-q really comes in handy here. Counting brackets is not the way or the light.
<rubic88>kmicu: The rollback feature is pretty amazing. Makes it much easier to experiment.
<another-user>nckx: yeah, doing that over ssh with bare bones emacs, sorry. need to install paredit :)
<nckx>Unbound variable: tar
<nckx>Unless you're getting different error messages than I am at this point, you now have a debuggable package ☺
<another-user>nckx: thank you for the help!
<kmicu>rubic88: thanks to first-class rollbacks messing with computers is a joy for me. At the same time updating my Debian NAS or OpenWRT router is like disarming a bomb in my experience 😺
<nckx>another-user: You're welcome. I hope you'll donate your package to Guix ;-)
*nckx spends a lot more time helping (well, trying to…) folks here than writing codez these days.
<another-user>nckx: i don't think guix will accept it because this packages doesn't build stuff from source but installs binary and building it from source is way over my head at the moment
<another-user>because this requires gradle and guix doesn't support that yet
<nckx>another-user: Ah. I'm not too familiar with ES or Java stuff in general (apart from its notoriety when it comes to bootstrappability). Pity.
<civodul>nckx: the time you and others spend helping is much appreciated i'm sure!
<another-user>civodul: i double that
<Marlin1113>hey guys
<Marlin1113>guix install: error: failed to connect to `/var/guix/daemon-socket/socket': Connection refused
<Marlin1113>getting this error
<Marlin1113>any tips?
<efraim>Looks like you need to start/restart your daemon
<efraim>are you on Guix System or Guix on a foreign distro?
<nly>Missing mtab file while determining whether /dev/SDA3 is mounted
<Marlin[m]><efraim "are you on Guix System or Guix o"> Foreing distro
<efraim>so that would be "sudo system guix-daemon restart" I believe
<efraim>it's been a while since I used something other than Guix System or systemd
<gduval>hi all, thanks to all contributors @GNU and Guix for all the great work! I've installed guix package manager on my debian system following the docs, installed emacs and emacs-guix. I had to manually add my home .guix-profile/bin directory to the path so that the right emacs gets launched. but then inside guix-installed emacs I get the following error when typing guix-all-packages : "error in evaluating guile expression : <unnamed port>:
<mfg>Hi, i've read the docs regarding the change of my default shell and i'm trying to find the correct expr for it. I want to add (shell <smth>) to my user account; this should be a gexp but i don't know what i am supposed to write. I want zsh to be my default shell.
<gduval>lookup : Unbound variable: package/output-sexps"
<cbaines>Is it possible to install Guix to a device from a system with Guix installed? I'm not sure init works in this case, as you'd need to do the cow-store step right? I think creating a disk-image works, but that's slower than I'd like as the image is built in a VM.
<gduval>so on my part emacs-guix is not usable
<efraim>cbaines: it should be possible to run "guix system init /mnt os-config.scm" without cow-store i think
<efraim>that's what I've tried in the past when I played around with aarch64 guix systems
<cbaines>I remember trying it in the past, and I think the command works, but I'm not sure the store on the target device has everything it should
<efraim>a disk-image would be another option
<cbaines>I'll perhaps try guix system init again...
<cbaines>mfg, something like (file-append zsh "/bin/zsh") I think
<cbaines>(which describes the bin/zsh file in the zsh package)
<gduval>any thoughts on how to remediate this?
<cbaines>gduval, it could be you're missing other environment variables
<cbaines>rather than setting PATH manually, try source ~/.guix-profile/etc/profile
<Marlin1113>the system guix-daemon restart doesn't work
<cbaines>Marlin1113, do you mean that restarting the guix-daemon service didn't work, or that you're still having problems connecting?
<Marlin1113>it's because there is no "system" command actually
<cbaines>What init system are you using?
<mfg>cbaines: thx, this works :)
<cbaines>mfg, great :)
<Marlin1113>shouldn't it be on /etc/init.d/?
<cbaines>Marlin1113, so, something like sudo service guix-daemon restart should do it?
<Marlin1113>unrecognized service
<gduval>@cbaines thanks for the hint, what kind of env variables?
<cbaines>Marlin1113, is the guix-daemon setup as a service on your system?
<cbaines>(as it sounds like it might not be)
<mfg>another thing i noticed is this: `guix system: error: symlink: Permission denied: "/var/guix/profiles/"' when running guix system reconfigure. Running with sudo gives warnings because root never ran guix pull.
<Marlin1113>dunno, i just executed the install script and it worked first
<mfg>Am i using guix wrong?
<Marlin1113>but after rebooting it doesn't
<Marlin1113>i'm running the script again
<Marlin1113>checking if i did anything wrong
<cbaines>gduval, I don't know a specific one you might be missing. I don't know much about the Emacs interface internals.
<gduval>cbaines thanks for the help anyway, your command yielded "~/.guix-profile/bin"
<cbaines>Marlin1113, the install script doesn't do any configuration in the case of sysvinit, you should have seen this message: unsupported init system; run the daemon manually
<bt`>Is there a clean and efficient way to do a filter-unfold-right with SRFI-1?
<cbaines>mfg, no, it's a bit unintuative. I've got in to the habit of using sudo -E when running guix reconfigure
<mfg>cbaines: Ah, i see. thx for the hint :)
<bt`>Also isn't this just (car host (aa (match (getaddrinfo host "nfs") ((x . _) x)))
<bt`>sorry (car (getaddrinfo host "nfs"))
<bt`>I guess what I'm really asking is if the error message difference between the match and the car would be relevant, and I'm guessing no.
<bt`>and that's for a specific case so just ignore me :)
<nly>You'd wanna use match (or something else)
<nly>I am not sure what type of data you're dealing with
<bt`>why's that nly? I'm just reading /gnu/build/file-system.scm
<nly>Car does not describe what you are trying to do (if you are looking for a specific item among unrelated things?) I think
<gduval>anyone else would have a hint as regards this emacs-guix error ""error in evaluating guile expression : <unnamed port>:21:16: In procedure module lookup : Unbound variable: package/output-sexps" ?
<nly>I am a newbie tho
<bt`>The only difference is the error message when you attempt to pass something other than a list/pair.
<bt`>that only matters if you handle the exception I guess, but I'm not familiar with how exceptions are handled in scheme.
<bt`>ah I misunderstood your comment then nly my bad. I'm going to be quite now.
<nly>No prob
<pkill9>hmm, i have an interesting thing happen, when i tried to build something with source, it put it in this path: /gnu/store/
<jackhill>Hi Guix: I want to express all the font-.* packages in a manifest, is there a good way to do that?
<pkill9>jackhill: I know there's a guile procedure that looks up packages (not sure if it supports regex though), if there is one for normal packages that supports that then you could use it
<pkill9>s/looks up packages/looks up inferior packages/
<pkill9>that procedure returns a list of package objects
<jackhill>pkill9: yeah. I guess I can also check what guix package --search uses
<pkill9>that procedure is called "lookup-inferior-packages" so there may be one called "lookup-packages"
*jackhill wil take a look, thanks
<pkill9>jackhill: ah, there's a comment in the source code for 'lookup-inferior-packages' that says "This is the counterpart of 'find-packages-by-name'."
<jackhill>although I guess that may not handle regular expressions
<pkill9>yea i just tested it, it doesn't :|
<jackhill>it looks like guix-package --list-available uses fold-available-packages with a custom funciton to do the regex matching
<joshuaBPMan>I just installed jami, but how do I start it?
<joshuaBPMan>ok gnome-ring. thanks
<divansantana>hi all :)
<divansantana>Trying a guix system reconfigure, results in [error] character map file `utf-8' not found: No such file or directory error
<pkill9>jackhill: i cargo-culted it out of guix/scripts/package.scm: but you sound like you understand it better than me
<pkill9>if you run `guile <that-snippet` it shows all the package objects found with the regex "^font-.*"
<pkill9>`guile <that-snippet>`*
<another-user>how do you deal with config files for packages?
<pkill9>copy them to the correct position in my home directory and modify them
<pkill9>or do you mean something else?
<pkill9>i like that guix system provides an API to the package manager
<pkill9>well guix, but it's out of the box in guix system
<another-user>pkill9: ah, i thought you're supposed to configure stuff via guix package configuration
<pkill9>another-user: you configure the system with a configuration file, and you configure the way packages are built using guix package configuration, which could include your own customisations, but there isn't a system for managing configurations of user applications
<pkill9>but all system stuff is configured in the system configuration, i.e. it generates a configuration file for things like ssh daemon
<pkill9>if someone or some people wanted to they could do the same with other applications, but it would be a lot of work
<pkill9>like all the configuration files for all the various packages you run as a user
<samplet>That will come eventually. There are a handful of motivated people who are excited for it.
<another-user>pkill9: got it, i'm building a package for elasticsearch and not sure what is the right way of dealing with its config file
<samplet>another-user: It should probably be configured via a “service”.
<samplet>Like “sshd”, etc.
<another-user>pkill9: so currently the only option is to manually create /etc/elasticsearch, right?
<another-user>samplet: could you please expand on that? looking at nginx package definity and don't see how they deal with config/logs
<pkill9>another-user: by default guix configures packages "/etc" to point to it's package path "/gnu/store/...-package", which is unmodifiable, so it won't look for config files in the system's /etc
<pkill9>unless they've hardcoded "/etc" in the code, i'm not familiar with elasticsearch
<pkill9>but packages in guix are built with the aim of not relying on the /etc/ directory, some have to for various reasons
<samplet>another-user: You will want to create an “elasticsearch-service-type” that tells Shepherd how to start Elasticsearch, which usually also includes generating a configuration file, telling Elasticsearch where it is, telling Elasticsearch where and how to log things, and so on.
<another-user>pkill9: samplet got it, thank you!
<samplet> the manual, and (as you know) there are a lot of examples in “gnu/services/...”. Good luck!
<bt`>of the three files I've looked over thus far I've seen a conditional-unfold and two filter-unfold-right, I wonder if a library with more powerful unfolds would be of use to the rest of the code base?
*civodul just watched
<civodul>i recommend it!
<samplet>Heh. I was at that talk. It was good. Deb has a really good perspective on things.
<civodul>yeah it's energizing and so clearly put
<civodul>i'm grateful Deb is spreading this message
*civodul does some testing on the hopefully last 1.0.1 installation images
<samplet>There was no Guix talk at LibrePlanet this year. :(
<civodul>it takes a while to download and install all these things
<civodul>heh, true, but there was a Guile talk, right?
<devilishtype>I hope it was guiless
<asterope>Is there any specific reason for grub's menu-entry definition to not have chainload capability?
<samplet>The Guile talk was good, but it felt... a bit early? I think the great Guile renaissance is still picking up steam. :)
<nckx>asterope: Only that nobody added it.
<devilishtype>what ever happened to guilemacs?
<civodul>samplet: early in the sense that Guile is still seen as "weird" and marginal?
<asterope>Oh, I need it, so maybe I'll try to patch it
<samplet>civodul: A bit, but more like the same talk in two years will have double or triple the number of cool projects to talk about. (I hope.)
<civodul>ah, sure
<civodul>definitely! :-)
<civodul>i feel like it's already much better than a couple of years back, but not yet as impressive as Rust or Go or Python ;-)
<civodul>we'll get there
<civodul>world domination! ;-)
<devilishtype>Impressive... Python...
*devilishtype gags
<jonsger>tbh, I feel like Guix is the project which "drives" most of Guile...
<samplet>Hooray! Gash just “successfully built /gnu/store/…-bash-boot0-4.4.23.drv” (with our bootstrap Guile).
<janneke>samplet: \o/
<samplet>janneke: I am hoping to release Gash-the-shell in a few days. Gash-the-suite-of-utilities needs a bit of work yet.
<janneke>samplet: we don't need the guile-2.0 compatibility anymore, right?
<samplet>Our bootstrap Guile is still 2.0.9 AFAIK.
<janneke>samplet: gash-theshell, that's great!
<samplet>There was some talk about updating it, and then a little bit of push back. It turns out that it was not so hard to port.
<janneke>samplet: i looked into who uses true,false,pwd as external commands a while ago; it's make-3.80 and old autotools (from glibc-2.2.5)
<mbakke>another-user: There are some elasticsearch patches on the issue tracker: <>.
<apteryx_>hi! which tool can I use to test dns resolution?
<apteryx_>I think there's something called nslookup, but from which package?
***apteryx_ is now known as apteryx
<apteryx>ah, bind:utils IIRC
<janas>Could someone else verify that emacs-idris-mode has an incorrect hash, and I'm not just going crazy?
<apteryx>it indeed mismatches
<another-user>mbakke: that's really great, thank you!
<janas>apteryx: thanks, I'll submit a patch
<nckx>apteryx: That also provides dig, which is better™.
<janas>Although I wonder how it got accepted, is it possible for packages in Melpa t change their contents without bumping the version number?
<apteryx>janas: probably. You could try looking up the snaphost of the original source (it must be cached by, and diffing its content against what's currently being pulled for that version from MELPA
<bt`>Is it generally safe to remove functions not used anywhere in the code base like this `make-skeletons-writable' or on occasion do individuals implement a function before they use it?
<apteryx>janas: I never remember the detail of the URLs involved, but you could look at this Scheme code, which automates such a thing for GitHub packages:
<jackhill>pkill9: this is what I came up with:
<apteryx>janas: this method actually will give you the url of emacs-idris-mode source archive substitute:
<rekado>g_bor[m]: it times out for me too (not the search itself, just the web interface in general). I’ll try to separate the web interface from the rest next.
<apteryx>jackhill: I just ran the procedure, and the URL of teh cache source should be:
<apteryx>but it gives a resource not found :-l
<apteryx>nckx: thanks! I keep forgetting about these utils.
<jackhill>apteryx: I guess above should be janas ☺
<apteryx>and they're hard to discover given the current bind description
<apteryx>jackhill: eeks. sorry.
<rekado>the story so far: hydra exists; hydra is often overloaded, web interface times out; cuirass is written; cuirass has no web interface; cuirass gets a web interface; web interface times out.
<jackhill>no worres, just wanted to make sure janas got flagged too
*rekado sneakily adjusts the timeout.
<apteryx>rekado: eh, good story :-p
<Marlin[m]>i gotta figure out why guix fails on my laptop
<Marlin[m]>It should be able to run libre
<Marlin[m]>I'm running devuan for now, but i'll investigate later
<civodul>at last! \o/
<Marlin[m]>Yay :D
<Marlin[m]>well, didn't even finish speaking and they fixed the bug
<civodul>see? :-)