IRC channel logs

2022-11-03.log

back to list of logs

<two[m]>(length (list-matches words 1))
<two[m]>s/1/str/
<vagrantc>well, patching guix/lint.scm to use (if (>= (string-length (string-replace-substring synopsis "@code{" "")) 80) fixed precisely one guix lint issue.
<vagrantc>i suspect the leading @code will get more, though
<vagrantc>that should fix more like 29 issues ... or at least change the issue
*vagrantc takes a moment to experience what little sunlight makes it through the clouds
*nckx zzz. No sun here. Good night, Guix.
<csepp>plopping this here as a teaser :) https://git.sr.ht/~raingloom/guix-source/log/mirageos-rebase
<csepp>the hello world example MirageOS unikernel builds with these packages :)
<pkill9>what is this?
<csepp>pkill9: mirage.io/
<pkill9>nice
<csepp>now off to sleep, i was just excited that i finally sorted my changes into individual commits and wanted to brag :D
<the_tubular> What's the package git:email ?
<the_tubular>I saw this on a git repo
***Xenguy_ is now known as Xenguy
***califax_ is now known as califax
<nanami>Do you mean git:send-email ?
<the_tubular>Yes, sorry nanami
<Korven[m]>So i'm using Guix on a foreign distro.
<Korven[m]>The distro had firefox pre-installed, and I decided to keep using it. But now, for some reason, it doesn't launch. If i try to launch from the terminal, it seems like it's relying on Guix's glibc and can't find __libc_pthread_init.
<Korven[m]>Any idea why this may have happened?
<iyzsong[m]>Korven: check env for LIBRARY_PATH, it should't be set
<Korven[m]>It isn't
<iyzsong[m]>um, does ldd /usr/bin/firefox has guix store paths?
<Korven[m]>it says it's not a dynamic executable
<Korven[m]>ok so looking into it, it doesn't have guix store paths
<Korven[m]>( i believe it was a symlink into /usr/bin/firefox )
<iyzsong[m]>okay.. you can try unset all guix envs (dont source its profile) and try, i think there are some env problems
<iyzsong[m]>you should have some thing like source $GUIX_PROFILE/... in your bash_profile, comment it and start a new login to check
<Korven[m]>okay, i will
<Korven[m]>yup, that does allow it to launch
<Korven[m]>it fails due to one of my extra profiles
<Korven[m]>just need to find out which one
<user_oreloznog>o/
<Korven[m]>So, more info.
<Korven[m]>I have 4 extra profiles. It looks like only when all 4 of them are sourced does firefox not work
<Korven[m]>which is very weird o.o
<user_>I still cant "guix pull"
<user_> https://paste.debian.net/1259330/
<user_>:( I ran guix gc --verify but not lucky
<vivien>user_, could you try and clone the official guix repo to disk, and then use a channels file (argument -C of pull) with the local clone? https://guix.gnu.org/manual/devel/en/guix.html#Using-a-Custom-Guix-Channel
<vivien>(I do that with an hourly cron job to sync it, so that I can still pull even if the guix website is down :D)
<user_>Sure, I should git clone "https://git.savannah.gnu.org/git/guix.git" ?
<vivien>Yes
<user_>clone is going it will tke some minutes then i just shoud use "guix pull -C path/to/guix" ?
<vivien>No, you have to create a channels file (the link should point you to the appropriate section)
<vivien>Just replace https://example.org/another-guix.git with "/some/where/with/guix" and the branch should be "master"
<vivien>But yes, cloning guix takes time
<vivien>Then guix pull -C channels-file.scm
<vivien>You can put everything in ~/.config/guix/channels.scm, but if you forget to undo this edit, then you will be cut off from any updates so this isn’t great :)
<user_>Unfortunately I sttill have the smae error
<user_> https://paste.debian.net/1259332/
<user_>those zeros at the and repet a lot of times
***devmsv_ is now known as devmsv
<vivien>OK
<vivien>Can you run guix pull as root?
<vivien>It won’t affect the system, but it will take some disk space
<user_>it give me the same error
<vivien>Then I don’t know :(
<user_>Im somekind lost this thing
<user_>I tried to downgrde and upgrade guix but still sme thing :(
<vivien>Are you in a foreign distro?
<vivien>If so, then you still have the option to remove guix, delete /var/guix and /gnu, and reinstall it…
<user_>Im at GuixSD :(
<user_>I think i cant remove it all
<user_>Maybe creating a new profile and trying to pull from the new one?
<vivien>That’s what you did as root I guess
<vivien>(or that’s what ended up happening)
<user_>I wonder if its possible to reinstall guix without formating the system itself
<vivien>What you could try to do is go to the clone of guix, run guix shell -D guix, and then compile guix, and ./pre-inst-env guix pull
<vivien>Maybe that will do something (I doubt it will but)
<Korven[m]>i'm realizing how annoying it can be to not have a working browser XD
<cbaines>the shepherd has started adding timestamps to the beginning of log lines, can that be turned off somehow?
<roptat>hi guix!
<user_>hey
<unmatched-paren>morning guix! :)
<unmatched-paren>lechner: Ah, I think you need to install bash-completion for that directory to be recognised
<nckx>...in the same profile, too.
<nckx>user_: --verify on its own would not do anything here. Did you use the suggested =contents,repair options?
<user_>Yes
<user_>I used both
<user_>still not fixed :(
<nckx>Well, poop.
<user_>the correct mail to aks more help is bug-guix@gnu.org ?
<nckx>To report bugs, yes. You'll be assigned a bug number etc. Questions are for help-guix but both are OK. Choose. :)
<nckx>Be sure to include the detailed output you shared yesterday.
<nckx>(Not as a link, those expire.)
<user_>Sent it, I also included the stack trace that appear after I issue guix pull
<Korven[m]>Is there any ways I can find which locations a store is symlinked into?
<Korven[m]>a store directory*
***Dynom_ is now known as Guest8420
<Obikawa>Hi. I have achieved my first package installable via a channel in my home configuration. I used a naked (package [code]) with guix package -f file.scm to iterate until success. After completing this process, I wish to able to start directly from (define-public package-name (package [code]))). However, I can't find documentation for this case. Is it
<Obikawa>possible to iterate on the full package definition without having to constantly wait for guix pull to fetch the next git commit?
<Obikawa>I think I found my answer after reading guix package --help again: guix package -L . -i package-name
<Obikawa>Yes - have now iterated enough to be sure that I found the correct answer.
<Kabouik>nckx I think a good part of the confusion about how much Guix is using on my SSD is coming from the fact that I started using btrfs for the first time when I installed Guix, and I am getting fooled by the usual tools I'd usually rely on to assess my disk usage, as they can't provide reliable estimates on a btrfs system (even though I am not using any RAID setup, I just have one SSD). My i3status-rust bar probably uses `df` too, and gives me
<Kabouik>pessimistic figures of the remaining space I've got. The btrfs tools are certainly more reliable there, though I'm still not comfortable interpreting them: https://0x0.st/og1m.txt
<Korven[m]>Kabouik: yo! i figured out a pretty sweet setup for reading pdfs and epubs :p
<Kabouik>Nice, using nov.el?
<Kabouik>I read my pdfs with mupdf
<Korven[m]>nope, zathura actually
<Korven[m]>I'm using Calibre + Calibredb (emacs) + zathura
<Korven[m]>so I can manage my library with Calibre(DB), and then emacs just uses xdg-open to pop it open in zathura
<Korven[m]>very minimal, quite nice
<Kabouik>I do have Zathura installed too but only has a backup application, I like mupdf's minimalism (but sometimes use Zathura if I want to extract an image for instance).
<Korven[m]>i just use mupdf as the zathura backend
<Korven[m]>now i'm just trying to figure out how guix messed with my Firefox
<Kabouik>Zathura's features remain mostly unexplored for me, I might move to it once I dive a little bit more into it and see if I can make it look as minimal as mupdf while knowing some useful :commands for it
<Kabouik>Heh, actually I can discover :commands by hitting : and then TAB, nice. Then I would just need to set some keybindings to use it instead of mupdf.
<Korven[m]>yes :V double the zathura user base
<Kabouik>Oh I'm pretty sure Zathura already has a quite massive user base :>
<Korven[m]>probably yes XD
<nikolar>Hello
<nikolar>Is there a POSIX compliant way to print out the permissions of the file in octal
***civodul` is now known as civodul
<Kabouik>Speaking of PDF management Korven[m], I'm trying to improve my academic literature workflow. I used to use Zotero before moving to Guix, which has some pretty handy downloaders for pretty much every journal (fetching metadata, filling the DB, and downloading/renaming the PDFs). Zotero is not available on Guix but I packaged CoBib, which is a TUI for Bibtex-based bibliographies, but I think it can only download from arXiv at the moment, and also I'm
<Kabouik>not using LaTeX yet, so citing literature from a Bibtex database is not exactly straightforward until I identify the right tools (either for LibreOffice, or moving to org-latex or LaTeX, but my co-authors won't).
***wielaard is now known as mjw
<Kabouik>I think there is a fair number of academic Guix users, but I assume the vast majority uses LaTeX and bibtex. Not sure if they have tools to ease populating their database and download PDFs though.
<Kabouik>I saw emacs-org-ref which seems nice, but it also can't download from all sources, and is obviously tied to using org-mode.
<Korven[m]>Well I'm very new to reading and writing academic literature myself, and I just use tectonic for writing my latex documents
<Korven[m]>although it takes a bit of work to make it work with AucTeX
<mothacehe>civodul: hey! I replied to your finalizer email :) The "if the wrapper disappears before the underlying C object" puzzles me. I may be missing something obvious. Regarding aggregation relations we are OK, afaict.
<civodul>mothacehe: ah good! lemme see
*civodul is working on installer warnings for unsupported hardware
<mothacehe>oh! great :)
<civodul>just replied
<civodul>i'm not diving too much into debugging this because my plate is already full of bugs
<civodul>i'll send the release update this afternoon
<nckx>nikolar: Not without sh scripting or using the POSIX-mandated C compiler AFAIK.
<Korven[m]>iyzsong: Just for completeness sake, after lots of debugging (aka logging in and out), I was able to narrow it down to what was causing the issue.
<Korven[m]>The culprit package is `papirus-icon-theme`. No idea why, but removing that allows firefox to run as usual
<Korven[m]>So it must be doing something that's messing with the paths
<nckx>Kabouik: Not an expert on how btrfs interacts with the Guix store, but I'll check if I see similar phenomena on the btrfs build farm. I'd be surprised, but that's half the fun.
<nikolar>nckx: well i am open to using scripts
*nckx AFK sorz.
<mothacehe>civodul: sure, thanks for replying, at least we are on the same page on the concepts :). I'll try to reproduce the crash with an instrumented installer with debug symbols this afternoon
<mothacehe>civodul: regarding the release update, we went through the 1.4.0 blocking bugs with apteryx on thuesday and demoted a couple of bugs as non blocking
<cbaines>civodul, also regarding the release, even though I managed to build hello locally for i586-gnu, there was some coreutils-minimal test failure when bordeaux.guix.gnu.org tried to build things
<cbaines>I pushed last night another change which should fix this. It's currently being built, so hopefully we'll see later today or tomorrow if it has worked
<elevenkb>Is there any way of specifying a luks2 partition according to its name instead of its UUID?
<elevenkb>In an operating system record.
<civodul>cbaines: yeah i saw that test-tls non-deterministic issue
<civodul>mothacehe: re dmemoting bugs, good! i'll take a look
<civodul>mothacehe: re Guile-Parted (and Guile-Git, which uses the same style), i think it'd be worth abstracting things a bit with a macro à la define-wrapped-pointer-type
<civodul>overall i feel there's too much repetition
<civodul>(that's orthogonal to the issue you're describing though)
<civodul>elevenkb: by "name" you mean the /dev/mapper/* name, right?
<civodul>that's definitely possible
<vivien>If I get some feedback for https://issues.guix.gnu.org/58932, I promise I will send the third try as 1 email per patch instead of attachments :) 🥕
<pkill9>out of random curiosity, does anyone use guix package manager on fedora?
<drakonis>run into selinux issues?
<civodul>vivien: heheh :-) note that if you send as one message per patch (with git send-email), you get automatic feedback via https://qa.guix.gnu.org!
<pkill9>drakonis: no I'm not using fedora, I was just wondering
<pkill9>I was wondering about the robustness between fedora and guix, since fedora uses a simpler system of just read-only images in the root filesystem, plus selinux
<pkill9>but downside is less provenance, maybe
<pkill9>and less customizability ofcourse
<pkill9>etc etc
<drakonis>ah right
<drakonis>silverblue and kinoite
<mothacehe>civodul: re define-wrapped-pointer-type: yeah you already proposed that in 2019, i'll take a close look once the Guile-Parted issue is fixed
<civodul>mothacehe: so i'm like a broken record, but i'm consistent :-)
<roptat>hi guix!
<roptat>I'm having troubles with android stuff... I'm trying to build their jemalloc fork, and it fails with an error that seems to be caused by these macros: https://android.googlesource.com/platform/external/jemalloc_new/+/refs/heads/master/include/jemalloc/internal/hooks.h
<roptat>src/prof.c:1414:8: error: reference to overloaded function could not be resolved; did you mean to call it?
<roptat>at this line: mfd = open(filename, O_RDONLY | O_CLOEXEC);
<roptat>so my understanding is that the macro expands to ((void)(hooks_libc_hook != NULL && (hooks_libc_hook(), 0)), open)(filename, O_RDONLY | O_CLOEXEC)
<roptat>and somehow the compiler doesn't know which of the four available "open" prototypes apply
<Korven[m]>Does anyone know where the guix src is located when installed on a foreign distro?
<roptat>somewhere in the store, but exact location varies
<roptat>what do you need the source for?
<Korven[m]>i wanna setup emacs geiser
<Korven[m]>roptat: i could have sworn i saw it in a .config or .guix-profile last time, just can't remember
<roptat>ah, there's a link from .config/guix/current I think
<Korven[m]>yeah! that
<Korven[m]>thanks :V
<Korven[m]>dang no autocompletions for module names
<roptat>ok, figured a workaround
<reyman>Hi guix guy !
<reyman>in my local dev project i need a mongodb database running, if i follow the guix way i don't need to pollute my home config and use a container.
<reyman>do you have some example of guix container example for database, mongodb-service-type for example ?
<roptat>reyman, hi!
<roptat>we don't have mongodb unfortunately
<roptat>and I'm afraid mongodb is not free software, so it can't be included
<roptat>I know it's not the answer you'd like to get, sorry :/
<roptat>the guix way would indeed be to use a container, I think there's an example of that in the cookbook, let me find it
<roptat> https://guix.gnu.org/cookbook/en/guix-cookbook.html#Guix-System-Containers
<stevenroose>So uhm, my Framework laptop doesn't seem to do suspension when lid closed or power button pressed. Is there some kind of GNU Hurd alternative for systemd suspend?
<stevenroose>Can't seem to find anything on suspending online
<examors>I don't know about the Framework specifically, but elogind can do suspend-on-lid-close with my laptop. I just had to set "(handle-lid-switch-external-power 'suspend)" in my config, otherwise it only worked when on battery.
<stevenroose>examors: aha, yeah I found loginctl suspend just now, so that works from CLI. I'll try find out where to put that elogind snippet. Thanks!
<cbaines>by the way, the issue pages linked to from https://qa.guix.gnu.org/patches now have a link to toggle the "moreinfo" tag of issues (via a mailto)
<cbaines>I'm currently using this for some issues that need a bit more work
<cbaines>e.g. when they've been reviewed, but need some changes before merging
<podiki[m]>very nice, thanks for all the great stuff with guix qa!
<examors>stevenroose: No problem :) The snippet should go in (elogind-configuration ...). If you're using %desktop-services then you'll have to use modify-services, something like this (untested) https://paste.debian.net/1259410/
<elevenkb>sorry for the interruption.
<stevenroose>examors: thanks! yeah I'm slowly starting to get used to the config.scm syntax :)
<elevenkb>To repeat my previous question: I've created a luks2 partition with cryptsetup luksFormat --label luksRoot.
<elevenkb>Is there any way to refer to the Luks partition in the mapped devices field of an operating-system record by using the name "luksRoot" as opposed to a UUID?
<reyman>@roptat thanks for you answer,
<Fare>How to I tell guix to tell grub to use lvm as well as cryptsetup, and then tell the kernel about the cryptsetup passphrase so I don't have to type it twice? Alternatively, how do I get guix to copy the /gnu/store files needed by grub to /boot ?
<reyman>i will try using mariadb or something else so ...
<apteryx>seems we should add -DBUILD_SHARED_LIBS=ON to our cmake invocation in the cmake-build-system: https://cmake.org/cmake/help/latest/variable/BUILD_SHARED_LIBS.html
<reyman>but @roptat, is it possible to only define postgresql-service-type into with-database.scm and load this as a profile to not pollute home.scm ?
<civodul>hi Fare! long time no see ;-)
<civodul>Fare: there's currently no way to do that
<civodul>the passphrase has to be typed twice
<roptat>reyman, you can add postgresql-server to a profile, but not a service, so you need a system declaration to run a container that can start the service
<roptat>that's what the cookbook is showing you :)
<civodul>Fare: LUKS on top of LVM should be possible though, see https://guix.gnu.org/manual/devel/en/html_node/Mapped-Devices.html#index-lvm_002ddevice_002dmapping
<podiki[m]>civodul: thanks for the suggestion on the emulate-fhs fix. your idea worked well and I have a test. not sure if what I have for both could be streamlined but I'll submit the patch today
<podiki[m]>I still need to think about how to handle --emulate-fhs with --profile or --manifest, because of the glibc-for-fhs package injection
<civodul>good!
<apteryx>podiki[m]: I've used -F to test a tox environment, and it was handy!
<abrenon>hey guix
<podiki[m]>apteryx: glad to hear it!
<civodul>howdy abrenon!
<roptat>hi abrenon :)
<elevenkb>civodul, all: I'll just refer to the partition as as /dev/sdax instead of trying to use the label. Wish me luck in installing.
<apteryx>does 'guix refresh --update' really need to recursively fetch submodules?
<apteryx>it takes ages on some repos
<apteryx>and this is typically frowned at in origins anyway
<apteryx>lilyp: about the emoticon in the new outputs list, I'd opt for a blank (nothing)
<abrenon>long time no see roptat
<roptat>:)
<civodul>apteryx: looking at (guix upstream), it's the job of git-reference->git-checkout, which clones recursively only when the <git-reference> object says to do so
<abrenon>roptat: would you happen to build maven projects with guix ?
<abrenon>I tried to follow the getting started guide in a shell with maven but it kept complaining about my lack of JAVA_HOME
<abrenon>can it be used as a standalone program or do I have to make a guix package ?
<roptat>you'll need a (openjdk|icedtea):jdk in the shell I think
<abrenon>yeah, of course
<abrenon>sorry I didn't mention that
<roptat>I don't think you should need anything else
<abrenon>I had openjdk:jdk
<roptat>let me try
<abrenon>thanks !
<roptat>mh, maven used to work, but I have the same error now
<roptat>indeed JAVA_HOME is not set
<podiki[m]>regarding #58549 and warning about hardware support, should there be a message or option to use nomodeset?
<podiki[m]>for me at least on newer amd hardware, using nomodeset let me boot and had graphics, though limited resolution
<podiki[m]>but totally fine to get set up
<abrenon>so it would be easier to simply make a mvn-build-system package out of a project to compile it ?
<podiki[m]>it is a common enough troubleshooting step that maybe it can be an option on the boot menu already
<roptat>abrenon, a simple workaround is to export JAVA_HOME=$GUIX_ENVIRONMENT I think
<roptat>probably the older versions I tried did not require a JAVA_HOME and could work without it
<apteryx>civodul: ah! then it's already doing the right thing. thanks for checking
<abrenon>I think I tried that last time, but now I fail to reproduce my own ^^
<Lumine>Good evening #guix
<abrenon>I found the exact I had run which was in a containered shell
<abrenon>I'm pretty sure last time even mvn --version crashed complaining about JAVA_HOME, but now it's returning me a version : S
<abrenon>now all I get is this error http://paste.debian.net/1259419/
<abrenon>I wonder why I no longer get the JAVA_HOME error
<abrenon>and why you get it now : )
<roptat>it might need access to your $HOME/.mvn
<roptat>$HOME/.m2 I mean
<abrenon>yeah, but there's none in my container
<abrenon>oh, .m2 ? didn't know about that one
<abrenon>neither exists in my ~, and I'm running from an empty directory in /tmp
<abrenon>trying again with icedtea
<roptat>that's where it'll download packages
<abrenon>oh, icedtea doesn't have a jshell : (
<abrenon>ok it's so working
<abrenon>the archetype:generate thinggy in the 5 mn tutorial is working
<roptat>:)
<roptat>arg, I'm creating loops between modules
<abrenon>and the example maven project I'm trying to compile is not stuck anymore
<roptat>good
<abrenon>it's doing something and failing at the end, but at least it's doing something
<abrenon>thanks for mentioning this alternative
<apteryx>abrenon: jshell was introduced with openjdk 9 or something
<abrenon>oh, icedtea is older ?
<apteryx>it's very old
<apteryx>we should hide it ;-)
<roptat>icedtea is for up to java 8
<abrenon>^^
<abrenon>I see
<abrenon>I get an error message with a "release 11" (I haven't understood release of what, but that seems to point in the same general direction)
<abrenon>Fatal error compiling: invalid target release: 11
<abrenon>I indeed get maven.compiler.source set to 11 in the pom file I was sent
<abrenon>ok so is maven in between icedtea and openjdk ? too old to understand the core libs provided by openjdk, but too recent to accept our old java version with icedtea ?
<roptat>you'll need openjdk >= 11 I think
<roptat>maven should be fine with any version of java
<roptat>but it needs a compiler that can target the source level
<abrenon>what does that mean ? could that be related to the error I pasted ?
<roptat>yes
<roptat>source level, is the Java version that the source was written for
<roptat>so 11 in your case
<roptat>and maven is looking for a compiler version that is greater than the source level, so it can compile it
<roptat>but you only have icedtea which is Java 8 at best
<roptat>can't work
<abrenon>that makes sense
<abrenon>now I still can't figure why maven doesn't seem to work with openjdk
<abrenon>I'm going AFK for a mo' but I'll be back
<roptat>ah maybe there's a compatibility issue indeed
<roptat>abrenon, you can try with openjdk@11:jdk, it seems to work better for me
<Korven[m]>Does Guix package JDTLS?
<roptat>"guix search jdtls" doesn't return any results
<Korven[m]>indeed, that's why I asked, if it's maybe under some other name
<Korven[m]>or if it's even packaged at all
<roptat>I don't know, never heard of it before, it's probably not packaged
<roptat>(at least jdtls would appear in the description I think, even if packaged under a different name)
<Korven[m]>:( sad
<roptat>I broke something, but I don't know what. Now trying to use a module in a guix repl -L . (where I'm in the sources of a channel), I get an error that something it defines and exports is unbound
<roptat>oh, it was a module cycle
<civodul>an article on how Fedora struggles with Rust packaging: https://lwn.net/SubscriberLink/912202/cf5ba56ea3863791/
<civodul>sounds familiar :-)
<florhizome[m]>Hi guix, someone willing to review https://issues.guix.gnu.org/58501?
<florhizome[m]>i think these are useful packages!
<florhizome[m]>Maybe I should create single tickets?
<unmatched-paren>florhizome[m]: for iio-sensor-proxy: the indentation in (arguments ...) is off, it should be (list #:key ...) not (list \n #:key ...)
<unmatched-paren>and the ``out'' variable should be replaced with #$output
<unmatched-paren>please explain why the tests are disabled
<unmatched-paren>and the configure-flags value should have a newline before it, as it spans multiple lines
<unmatched-paren>synopsis should just be something like "Proxy sensor devices to applications through D-Bus"
<peterpolidoro>anyone here use mbsync to fetch email? I am trying to get it to work with xoauth2
<unmatched-paren>and description: "@code{iio-sensor-proxy} proxies sensors like accelerometers, light sensors, and compasses to applications through D-Bus"
<unmatched-paren>s/D-Bus/D-Bus./
<unmatched-paren>home-page should go before synopsis
<unmatched-paren>and sysopsis has one space before it instead of two
<unmatched-paren>the (native-inputs ...) line is too long
<unmatched-paren>s/line/lines/,s/is/are/
<unmatched-paren>for power-profiles-daemon: please use the gexp arguments form
<unmatched-paren>and remember, (assoc-ref outputs "out") is obsolete; use #$output
<tribals>Hi, folks! It's me again
<unmatched-paren>home-page should go before synopsis, and synopis should not have a period after it
<unmatched-paren>the description is way too long imo
<unmatched-paren>for low-memory-monitor:
<florhizome[m]>do you think it’s the right module?
<abrenon>roptat: brilliant piece of advice, thanks again !!
<abrenon>the demo software I was sent now compiles
<unmatched-paren>use gexp arguments there, and put home-page before synopsis
<unmatched-paren>it might be nicer to use (list glib "bin") instead of `(,glib "bin")
<unmatched-paren>(for all three
<unmatched-paren>)
<unmatched-paren>florhizome[m]: yeah, that's the right module
<unmatched-paren>and that's all i can see
<tribals>I've tried to pass two `--listen` options to `guix publish`, and unfortunately, it does not work. As you can remember, I tried to do this because I want `guix publish` to listen on both IP v4 and v6 interfaces. Currently, I can make it so only by abusing `net.ipv6.bindv6only` sysctl. But I'm planing to switch this on, as I mostly agree with those considerations: https://utcc.utoronto.ca/~cks/space/blog/linux/Bindv6onlyNote
<tribals>This is mostly no-action reminder to anyone interested.
<elevenkb>Hello there when I try to install guix onto my system I get an error involving keyboard-layout.
<unmatched-paren>elevenkb: hello! could you please put the error on paste.debian.net?
<elevenkb>unmatched-paren: I'm in the install disk without a graphical environment so I don't know how to submit the error there.
<elevenkb>I can redirect the output of guix system init to a file.
***mark_ is now known as mjw
<elevenkb>What to do from there?
<elevenkb>Nevermind I probable see where the error is coming from
<unmatched-paren>elevenkb: hmm, i'm pretty sure there's a program you can use with ``guix shell'' for submitting to some paste site...
<unmatched-paren>elevenkb: try ``guix shell wgetpaste''
<roptat>wgetpaste?
<roptat>:)
<elevenkb>unmatched-paren: don't worry about me for at least 10 minutes. Essentially instead of one operating-system record definition I split up my desktop.scm into multiple definitions for clarity.
<elevenkb>So the entry (keyboard-layout keyboard-layout) can't be extracted from the operating-system definition because the variable in second position is out of scope.
<florhizome[m]><unmatched-paren> "and remember, (assoc-ref outputs..." <- guess i don’t need lambda* to preserve #:key outputs then ?
<unmatched-paren>florhizome[m]: nope
<unmatched-paren>you can just use (lambda _ ...)
<podiki[m]>yup, is a bit simpler now, once you get the hang of it
<unmatched-paren>yes, and you can use #$output:doc for the "doc" output, #$output:foo for the "foo" output, etc
<unmatched-paren>same for gexped packages, with #$pkg:foo, etc
<unmatched-paren>i do wonder whether you can do something like #$(this-package-input "foo"):out or something
<florhizome[m]>nice
<podiki[m]>I don't think so, wouldn't #$(...) give you a string?
<podiki[m]>so you'd get "/gnu/store...":out but I don't know the reader order here
<unmatched-paren>podiki[m]: there might be a special case in the reader macro for ungexp-followed-by-colon-with-word
<podiki[m]>right
<podiki[m]>did you try it? :-)
<unmatched-paren>nope :)
<podiki[m]>exercise for the reader
<podiki[m]>pun was not intended, but I'll take it
<elevenkb>I have
<elevenkb>solved the issue with keyboard layout. Now on to the next one:
<elevenkb>When I try guix system init desktop.scm /mnt an error shows up: "No space left on device"
<unmatched-paren>how big is the /mnt device...?
<elevenkb>460.4G?
<unmatched-paren>you have a 4GiB hard drive...?
<abrenon>and how much free space does it have ?
<elevenkb>G Stands in for GB
<elevenkb>In this case.
<unmatched-paren>elevenkb: i know, but i don't think that's your hard drive....
<unmatched-paren>did you mount the USB on /mnt?
<unmatched-paren>it should be the hard drive
<jab>hey guix! Has anyone tried guix via multiseat?
***liberalartist is now known as philip1
<podiki[m]>for anyone using --emulate-fhs this patch should fix the bug with the wrong glibc ending up in the container: https://issues.guix.gnu.org/58861#4
<podiki[m]>(so if you needed LD_LIBRARY_PATH before for e.g. /lib, shouldn't now)
<Fare>civodul: thanks
<abrenon>well see you folks
<abrenon>thanks again for the great tips !
***Awoobis is now known as A_Dragon
***jesopo is now known as jess
<Kabouik>Any ytfzf user here? I installed the package from Guix (I am not using the script directly) but every search ends up with "marker display width should be up to 2", in any terminal emulator. I couldn't find that sentence in the ytfzf sources, weirdly.
<elevenkb>Yah I don't think I'm mounting my USB drive as /mnt.
<elevenkb>When I try guix pull --chabnels
<elevenkb>Sorry, when I run ``guix pull" I get an error: "cannot expand packfile /root/.cache/guix.../pack_git2_9ffFsi."
<elevenkb>df -h shows that the volumes mounted under /mnt all have at least 96% free space.
<elevenkb>So that can't be the issue.
<SeerLite>Hi! package-input-rewriting requires rebuilding all dependant packages, right? Is there a similar procedure that grafts instead?
<podiki[m]>There's with-graft
<podiki[m]>I don't know the scheme procedure off hand, but that's the guix build option
<apteryx>hm, why does calling "(set-build-options store #:offload? #f)" twice in a row exits with an exception?
<apteryx>where store is (define store (open-connection))
<lilyp>does (set-build-options) close the connection by chance?
<Kabouik>jgart[m]: I see that you were involved in the initial patch to include ytfzf, does it still work on your end? I'm getting this weird "marker display width should be up to 2" issue that I can't find in any web search or in the ytfzf source. I tried both the package and directly the ytfzf script.
<jackhill>Kabouik: basic functionality seems to work for me at guix commit e26b529b1ab4671a971f0a4b45d19abe55134475
<f3n1x>heya guixers !
<Kabouik>Hum, interesting jackhill. For me it fails with any terminal emulator. Maybe because I'm using Wayland?
<Kabouik>I might just move to emacs-ivy-youtube, but that requires a Google account, while ytfzf doesn't and can still manage subscriptions with just a list of channel urls, which is quite nice.
<f3n1x>i'm starting to use a brand new '2ary pc', how can 'i clone ' the system configuration from my laptop to this '2ary pc' ? May i just export my current /etc/configuration.scm to ? Thanks, thanks, thanks
<jackhill>Kabouik: I'm useing foot with sway, so who knows!
<Kabouik>I'm using foot with Sway as well. The plot thickens.
<Kabouik>I just set up emacs-ivy-youtube as a plan B and it's quite nice, but that it requires an account annoys me deeply.
<nckx>It needed kickment.
<nckx>f3n1x: Basically. You'll have to partition (& mkfs &c.) the machine before installing Guix System, but the rest is defined in your .scm(s).
<nckx>If you're not too familiar with doing that by hand, you can always install a bare-bones Guix System just using the graphical installer (creating the same user, but no fancy packages/services), reboot into it, then copy your system configuration to /etc and ‘sudo guix system reconfigure’ to transform it into home.
<nckx>Then you have a known-booting system to roll back to if needed (say, your disc layout wasn't quite the same…).
<f3n1x> thX,thX, thX nckx
<SeerLite>podiki[m]: Thank you! I misread the manual and didn't realize transformations also do recursive replacing. With that I can just use options->transformation from (guix transformations) :)
<rekado_>nckx: uh, did goggles-bot die?
<nckx> https://logs.guix.gnu.org/guix/2022-11-03.log#233745
<nckx>No.
<vagrantc>goggles-bot: botsnack
<nckx>rekado_: Is something missing?
<rekado_>oh, I guess I just misunderstood then
<rekado_>“it needed kickment” sounds like it fell asleep or died and needed a restart or something
<nckx>Oh no no no, nothing so bad. I touched today's log file by mistake and it doesn't like that, and stopped logging, so I had to kick it.
<rekado_>I see
<rekado_>thanks for restoring the missing logs
<nckx>(‘By mistake’ = I guess goggles logs slightly less than ZNC(?) and I relied on wc -l at first.)
<nckx>rekado_: Is there anything else still missing besides a proper service?
<rekado_>I think it’s working fine, but I’m not at all confident about the weird mechanism to close and start a new log file.
<rekado_>it would not surprise me if I overlooked something there and merely built an accidentally useful collection of bugs there.
<rekado_>I know very little about IRC, so maybe the filtering is a bit too aggressive.
<rekado_>it only logs messages that were sent to the channel according to guile-irc’s parse-target.
<nckx>It probably just duplicates some filtering previously done by goggles proper, which is arguably more correct anyway.
<nckx>And we have plenty of distributed back-ups if a data-eating bug were found.