IRC channel logs


back to list of logs

<magnumDK>yea manual was no help..
<ngz>mirai: It creates an appropriate texlive-updmap.cfg according to some default packages (including texlive-bin) and optionally provided ones. It is not an union of directories.
<ngz>mirai: It merely propagates its arguments.
<nckx>magnumDK: We'll need more info than that.
<chipb>amjoseph: oh no are you jumping ship to guix? ;-P
<nanounanue>Hi everyone
<Gooberpatrol66>it feels so good being on KDE
<Gooberpatrol66>i'll never have to deal with the gtk file picker again
<Gooberpatrol66>brings a tear to my eye
<nanounanue>I have a question related about how to use the docker package under guix. I installed the docker package in my guix home. When I run it, e.g. using docker ps, I got the following error: "Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?" What I should do?
<PotentialUser-76>nanounanue That means the docker daemon isn't running usually. You have to start it somehow (I'm a guix noob, so I don't know how, but there must be a shepherd command for it)
<PotentialUser-76>Also you could try rebooting. Sometimes packages configure the docker service to start automatically at next boot.
<nanounanue>Thanks @PotentialUser-76, I didn't find any example of a shepherd command for it :(
<PotentialUser-76>Hi all, I've never used guix before. I'm thinking about migrating to it from Arch because I dislike systemd, dislike having to constantly update everything and dislike how people on arch forums are not very nice. I'm trying to figure out whether guix would be a good fit for me.
<nanounanue>I am a happy user of guix sinde two years ago, and I am very happy with it, highly recommended
<PotentialUser-76>Btw I was trying to search for how to activate the service and I found a ChatGPT post lol:
<PotentialUser-76>(those instruction are nonsense btw in case it's not obvious)
<nanounanue>PotentualUser-76 LOL :)
<PotentialUser-76>nanounanue Well one thing I'm wondering is obviously guix doesn't have as many packages as Arch/AUR. If I want to install something but there isn't a package for it, what happens then? AUR for example is nice in that any rando on the net can just upload their shell script, so even though not all AUR packages are good at least you can almost always
<PotentialUser-76>find scripts there.
<PotentialUser-76>Oh btw can you try running `dockerd`? That will run it manually in the foreground: While it's running, try docker ps again in another terminal window
<nanounanue>Well, the standard answer is that you can create your own package definition. There are several "build" systems that will help you to, well, built the package that you want. Another easy option is that "import" a package from cran, pypi, etc:
<nanounanue>PotentialUser-76: dockerd not found
<PotentialUser-76>can you try to search for it in root with find or fd?
<PotentialUser-76>For me it is in /usr/bin/dockerd
<PotentialUser-76>Oh that import sounds really neat
<PotentialUser-76>I guess I'm not the first person to wonder about importing from arch packages or aur
<PotentialUser-76>So my biggest concern with guix is getting my hardware to work. Beyond that, it sounds like everything else would be straightforward.
<PotentialUser-76>But my partition table is a bit complicated and I don't want to resize them. But if I run guix in a vm it won't tell me if the hardware is compatible, since it would use the VM's fake hardware.
<PotentialUser-76>Hmm... Can I install guix in a chroot, and try out the drivers that way?
<nanounanue>Mmmm don't know, but you have some options: You can keep Arch as your "base" OS and use guix as package manager
<nanounanue>I am using it in that way (with Ubuntu)
<nanounanue>That option is called "Guix in foreign distro"
<PotentialUser-76>Yeah, that would allow me to try out the package management
<PotentialUser-76>But not eg. graphics drivers
<PotentialUser-76>I think the live usb is supposed flag unsupported hardware, so maybe I'll try that
<lilyp>PotentialUser-76: quick question, what hardware are we talking about?
<lilyp>if it's an intel cpu with on-board graphics, it'll probably work
<lilyp>for amd, it's trickier
<PotentialUser-76>Well, I don't know what would break until I try it. But I'm guessing wifi and gpu will be a problem. It's an nvidia card.
<lilyp>are you currently using nouveau on arch?
<PotentialUser-76>No, the proprietary driver
<lilyp>that might indeed be an issue then
<hwpplayer1>hi GNU Guix People !
<sughosha>Hi, is there a way to have entire system configure with guix transformation? I want to transform all the services with a graft.
<nanounanue>I have a question related about how to use the docker package under guix. I installed the docker
<nanounanue> package in my guix home. When I run it, e.g. using docker ps, I got the following error: "Cannot
<nanounanue> connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?" What I
<nanounanue> should do?
<jpoiret>nanounanue: you need to have a docker daemon started
<nanounanue>(I apologize for the weird format)
<jpoiret>you're on a foreign distro so you need to have the docker daemon running with your host distro i think
<jpoiret>or just manually start the docker daemon in the guix package with the proper permissions
<nanounanue>jpoiret: So I need to install `docker` and `docker-cli` right?
<nanounanue>(because I only have `docker-cli` right now)
<jpoiret>not just install the packages, but also properly start the docker daemon
<nanounanue>Ok, thanks
<nanounanue> what do you mean by "properly start the docker daemon"?
<jpoiret>well, you have to start the daemon under the right user, with the proper config and all
<jpoiret>something that's usually handled by systemd/shepherd services
<Altadil>Hi, I have a noob question: I’m trying to update a package with guix refresh. guix lint reports a problem, but it’s already in the current version. Am I correct in thinking the update and correction should be two different patches ?
<next4th>Altadil: yes, that should be 2 patches
<Altadil>next4th: thanks!
<xelxebar>Gah, trying to package arrayfire and one of the build steps is wanting to clone some upstream repos or something.
<xelxebar>Hopefully there's some CMake variable to point it at a local checkout.
<lilyp>ACTION 🤝️ hako [Toaruverse]
<minima>hi, i have this package definition that doesn't build - it fails the check phase; how do i go at debugging this? i do see a build log and the build log says that `' is failing, but how do i take it from there?
<minima>do i manually download the package source code and launch the tests from there?
<minima>is there any facility in guix that semi-automates any of this?
<rrobin_>minima: you can call 'guix build -K -f pkgfile' to build it and it will keep the failed build tree under /tmp for you
<rrobin_>you can retry the build manually after loading environment-variables
<rrobin_>see here
<nckx>Once in the build directory, you can use ‘guix shell -Df’ to *approximate* the build environment.
<minima>ah! that's brilliant, thank you rrobin and nckx!!
<stikonas>who managed letsencrypt certs here?
<stikonas>I think autorenewal is broken and guix cert is about to expire
<stikonas>only 2 days left now on
<minima>with `guix shell', is there any specific expected interaction between `--container' and `--development'
<xelxebar>!@#%!#$% Knee deep in CMake hell. Unable to package ArrayFire but really need it for a project.
<minima>by using both options i was expecting to get the build dependencies of a given package, within a container
<xelxebar>Willing to pay someone to package it if there are any takers.
<minima>however, it seems that the package itself is being built?
<xelxebar>Here's what I currently have:
<minima>oh... wait... `--development --container' is not `--container --development'... ouch
<minima>they behave differently
<minima>is that expected?
<xelxebar>next4th: You interested? :P
<next4th>xelxebar: well, honestly not much for me, gpu/ai/ml things is scaring lol. you should send it to guix-patches
<roptat>hi guix!
<xelxebar>next4th: Ah, okay. Guess we're kind of in the same position, then. hehe
<minima>(oh sure, `--development --container foobar' is clearly different from `--container --development foobar' and they way i wanted it is the latter, where the package you want to work on comes after `--development')
<minima>*the way
<next4th>minima: yes, maybe it should report a error when '--development' is not followed by a package
<stikonas>roptat: do you have the rights to check what's going on with autorenewal of letsencrypt cert on ?
<stikonas>the cert will expire in 2 days
<roptat>stikonas, I don't have access to the servers
<renngar>I have an error with a module loading at boot. How can I change the current-module-debugging-port contained in linux-modules.scm, so that the "loading Linux module" messages display so I can see what is failing?
<stikonas>oh, I see. Do you know who might?
<efraim>stikonas: I think I renewed it, can you check?
<stikonas>efraim: not yet, still getting one. Have you reloaded web server?
<stikonas>efraim: do we not have any autorenewal hooks?
<efraim>not yet, I'll check the guix-sysadmin mailing list to see if there are better instructions than what I guessed
<efraim>there are automatic mcron jobs to do it, but they sometimes fail
<stikonas>also might benefit from renewal too...
<stikonas>less than a month remains
<stikonas>and I think it's also running on the same guix infra
<mirai> <>
<mirai>re certbot troubles
<roptat>is there a way to debug the shepherd? I just updated my system and it becomes unresponsive quickly after boot (~1 min)
<roptat>when I boot the system, I can run "herd status", it tells me openssh and smtpd have failed to start, but I don't get network before 1 or 2 minutes after boot, and at this time even "herd status" is unresponsive
<mirai>so its not the “shepherd can't reboot or shutdown” that occurs after a shepherd gets upgraded to a higher version with a reconfigure?
<roptat>no, it's even after a cold boot
<mirai>what shepherd version is this
<roptat>it's on aarch64, the latest (pulled today)
<mirai>are you using network-manager ?
<mirai>to provide network
<roptat>no, dhclient
<roptat>it's a small SBC
<mirai>hmm… whats the time on the system
<mirai>be it RTC or whatever the system is using
<roptat>initially, 0, then at some point the time is correct
<roptat>I can see in the logs ntpd is setting the time once it has network connectivity
<mirai>maybe <> ?
<roptat>ah, maybe
<roptat>is there a way to set a specific time at boot? the SBC doesn't have a battery to save time
<mirai>I'd say having a battery backed RTC would be the better route (something like a DS1307 module perhaps?)
<mirai>but you could do something similar to the urandom seed service
<mirai>if the board isn't powered off too long, maybe it can work?
<mirai>ngz: when exactly should I use texlive-updmap.cfg? Is it necessary for the dblatex package definition?
<user_rngh> hi all! i'm trying to install guix 1.4.0 from usb-stick (i use example config from /etc/configuration/desktop.scm) and installation fails in python-afdko-3.9.1 (check phase). i tried to find a solution - no luck. any advice? this is my first meet with guix.
<roptat>user_rngh, one possibility would be to try the latest instad of stable:
<roptat>*latest instead of standard
<roptat>if you could share the content of /var/log/guix/drvs/x4/z372iblc6dci0sgfrzzn0kngl91pn0-python-afdko-3.9.1.drv.gz (once decompressed) with us, it would be nice :)
<user_rngh>i cant get latest image by some reason...
<roptat>so another solution would be to install a bare-bones system first, run guix pull and reconfigure a full desktop system from there
<roptat>the first system would be less likely to fail building
<user_rngh>i already try simplified config. dwl instead of xfce+gnome
<roptat>I mean, without graphics, there's an example of a bare-bones system:
<roptat>I don't think it would need afdko
<roptat>once it's built, you boot it, run "guix pull" to get the latest version, and reconfigure with the config for your graphics system
<user_rngh> graphic at all. got it
<user_rngh>thx i will try
<roptat>good luck :)
<user_rngh>so...i cant get latest packages from 1.4.0 iso immediately? i read  guix manual and didn't see any related options
<user_rngh>...or i miss something?
<roptat>mh, you could run "guix pull" to get the latest, but I think it would get it in the tmpfs, so it will require a bit of RAM
<roptat>well no, not a tmpfs, I'm just confused. I'm just not sure it will work since the file system might be a bit small for that
<roptat>you could still try
<roptat>(after "guix pull" make sure "guix describe" tells you it's the guix you just updated, and not the previous version)
<user_rngh>i tried actually...and messing up with system on usb-stick. some how it try to install unrequited packages like zenity and gnome parts after "guix pull" and "guix system init /mnt/etc/my_minimal_config.scm /mnt". i probably do something wrong previously.
<user_rngh>i remember that. thank you :)
<PotentialUser-59>Hey all, I'm thinking of moving to Guix from Arch. I'm curious about updates. Arch is really annoying with how it expects to constantly do full system updates (like every day) which take a lot of time and sometimes break things. People on the Arch forums also get really mad if you do partial upgrades. Guix sounds like partial should be a non-issue,
<PotentialUser-59>but it does say that it's rolling release. So is the Guix community very opposed to partial upgrades?
<ngz>mirai: texlive-updmap.cfg is mandatory to take into consideration fonts installed along the inputs in the package. Otherwise, fonts are installed when updating profiles, which does not happen for fonts in (native-)inputs.
<mirai>ngz: would it make sense for eso-pic to propagate texlive-graphics? This is one of the lines in the log: /gnu/store/xm7db2x02pfk2qas2kb1l9kif85549g4-texlive-eso-pic-66594/share/texmf-dist/tex/latex/eso-pic/eso-pic.sty: File `keyval.sty' not found.
<ngz>mirai: Nope, for sanity reasons I blindly follow "texlive.tlpdb" recommendations for propagation. You simply need to add both to inputs.
<ngz>mirai: It only matters for package developers, tho. As a user, you probably want to install whole package collections anyway.
<ngz>mirai: For example, texlive-eso-pic is located in texlive-collection-latexrecommended, which, in turn, propagates texlive-collection-latex, contaning texlive-graphics. IOW, installing texlive-collection-latexrecommended would install both.
<janneke>PotentialUser-59: what's a partial upgrade?
<janneke>fwiw, /me doesn't update all that often
<PotentialUser-59>janneke When you update only one package instead of the whole system
<mirai>ngz: tbh I'm not so sure what to do about dblatex. I can continue hunting down what's left to get the test/sample file to build but maybe the whole thing should be included instead? I've tweaked the package to this <>
<mirai>dblatex3-0.3.12/docs/install.xml says this
<mirai>Install the minimal distribution, and add the following packages: changebar, colortbl, fancybox, fancyhdr, fancyvrb, listings, overpics, rotating, subfigure, titlesec, bibtopic, enumitem, eepic, lm, lastpage, helvetic, times, symbol, courier, footmisc, ifxetex, pdfpages, wasysym.
<ngz>mirai: What sample file? I got the one from yesterday built.
<mirai>ngz: sample file being the test one, yes
<ngz>mirai: Note that texlive-updmap.cfg brings all packages from texlive-collection-basic, so you should not need to add those manually (except for texlive-bin, for $PATH issues, AFAIU).
<ngz>mirai: So, what is left to do?
<mirai>Try to zcat the log since it doesn't yet build, it complains about missing things
<mirai>some bash helpers for this: <>
<mirai>ngz: I haven't finished enumerating the texlive-packages it complains about yet
<ngz>mirai: Do you have a log file to show?
<janneke>PotentialUser-59: Ok; guix works a bit differently i guess. Usually you would update a profile with packages to new commit that defines some set of package versions. Updating your system profile and your user profile(s) need not to be done simultaneiously
<mirai>ngz: <> (note, I've added more texlive packages in the meantime)
<nckx>PotentialUser-59: There are edge cases where you must update certain packages together, but that's not Guix's choice. Guix makes partial updates as possible as possible.
<PotentialUser-59>oh well that sounds great! that's what I was hoping to hear :)
<ngz>mirai: You need texlive-ec, texlive-zapfding…
<ngz>mirai: texlive-symbol
<ngz>mirai: But the fatal error is that you need to add texlive-intcalc.
<ngz>mirai: I think the rest is pdflatex being verbose.
<ngz>mirai: I need to go AFK for some time, if you send me logs, I can help you find what packages are missing.
<mirai>will do once I sift the fatal ones out
<Guest28>How can I give libpulse as input for a package?
<Guest28>ah nvm, forgot pkg-config
<ngz>mirai: back
<mirai>ngz: I've mailed you the logs & co.
<ngz>mirai: You don't need to add TeX Live inputs twice. texlive-updmap.cfg already propagates them, so they should appear as regular inputs during dblatex build process.
<mirai>I've got to run now but I'll reply asap (here or mail)
<mirai>ngz: its to set GUIX_TEXMF in the wrapper
<ngz>I'm talking about `texlive-packages' in inputs.
<ngz>This is orthogonal to the wrapper part.
<ngz>mirai: So, in a nutshell, you can remove `texlive-packages' variable from input (and replace `cons*' with `list', obviously). You also need to remove (unset "GUIX_TEXMF") from `check-wrap'. Furthermore, the GUIX_TEXMF dance in `wrap-dblatex' is apparently not needed, so you could drop the whole texlive-packages binding and simply add this list as an argument to texlive-updmap.cfg.
<attila_lendvai>anyone from the python crew around?
<radio->I am dealing with a pretty strange situation...
<radio->Seems that I'm unable to create a shell with texlive-scheme-minimal or texlive-scheme-basic
<radio->despite being explicitely using the (gnu packages tex) module
<radio->and both of these symbols being exported by that module
<radio->(@@ (gnu packages tex) texlive-scheme-minimal) also doesn't work
<radio->And using specifications->manifest instead of packages->manifest don't work either
<radio->Someone here have a guess of what could it be?
<radio->Found my mistake
<PotentialUser-78>Hi, is it intentional that signal-desktop for signal messenger is not included as a package on guix? Are there other signal clients that are packaged?
<cmmm>Hello! I'm trying to setup emacs-geiser for my guix system
<cmmm>configuration, but there is one problem. Geiser has issues finding
<cmmm>modules from the nonguix project, which I download as a channel (note
<cmmm>that ~# guix system reconfigure ...~ works just fine).
<cmmm>I had no luck finding where guix clones channels, and there seems to be no nonguix package, that would add it to the GUILE_LOAD_PATH as a library. The other option is to clone nonguix into my home directory, but it's not that clean, because I'll have to pull it, when updating my channels to keep it up to date. Any ideas how to do this in a clean way?
<roptat>PotentialUser-78, I'm not sure if there are freedom issues with it, but at least, it's an electron app, which is extremely hard to build for us
<PotentialUser-78>roptat, thanks. Do you think electron apps will be difficult to build for the forseeable future? Or, once it is "figured out" for one electron app, then it will be easier for subsequent ones?
<roptat>once it's figured out for one, it'll be easier for others I think
<PotentialUser-78>Okay, thank you
<roptat>cmmm, I don't know geiser, so I'm not sure that's what you're looking for, but you should be able to find a repository in ~/.cache/guix for each of your channels. Or maybe you're looking for ~/.config/guix/current instead?
<cmmm>roptat: Thanks! The directory ~/.config/guix/current/share/guile/site/3.0/ does just the thing, but it's kinda weird, that it's in ~/.config
<roptat>it's actually a symlink to /var/guix and then to some place in /gnu/store
<nckx>The symlink 'target' is 'configuration', if that makes sense. And it does, once you look at it that way 😉
<renngar>Using 'guix system vm' I found out that dm-crypt is the module failing to load. It and all of its dependents are in the directory, but I have no idea why it is failing to load.