IRC channel logs


back to list of logs

<dftxbs3e>mbakke, also not working on ungoogled-chromium:
<bone-baboon>I am trying to install Guix stable 1.2.0 x86_64 release. When I boot it all I get is a gray background with the Guix logo. There is nothing else. No graphical installer. When I try C-A-<F1-F12> nothing happens I am stuck at the gray background with the Guix logo.
<dftxbs3e>bone-baboon, no error logs whatsoever during boot?
<dftxbs3e>what grey background exactly? can you take a picture?
<vagrantc>what video card? there have been problems with some video cards... i don't know much in the details
<dftxbs3e>vagrantc, the installer is still text based, no?
<Noclip>Neither snapper nor timeshift are available in the guix repository. How can one do automated snapshots on btrfs?
<dftxbs3e>Noclip, I can package either
<dftxbs3e>Noclip, but GNU Guix has capabilities allowing system configuration rollback and forward smoothly, unless you want data as well
<dftxbs3e>Noclip, Timeshift says: "Timeshift is designed to protect system files and settings. It is NOT a backup tool and is not meant to protect user data." - GNU Guix already handles it on its own.
<Noclip>dftxbs3e: I don't need them currently, I was just wondering that they aren't available while guix seems to have special support for btrfs.
<vagrantc>dftxbs3e: yes, but even on the console some video cards cause issues
<dftxbs3e>vagrantc, okay, I guess then bone-baboon you should be able to modify the grub command line to disable your graphics card's driver
<Noclip>But if you have the time and motivation to package them feel free to do it. Sooner or later someone might need them.
<dftxbs3e>Noclip, when it comes to handling system/config rollback though GNU Guix handles it conceptually already, it may only be useful for user data
<dftxbs3e>"Thanks to Snapper, you can mess up system configuration changes or package installations or updates without having to restore from an old backup and risking to lose some files. Just revert to the snapshot before your problematic change and you’re fine." - it also looks to be snapper's primary use case, again, not needed on GNU Guix.
<Noclip>"it may only be useful for user data" That's what I was mainly thinking about.
<dftxbs3e>" The standard setup of Snapper is designed to allow rolling back system changes. However, you can also use it to create on-disk backups of user data." -
<dftxbs3e>let's package snapper then
<Noclip>I assume the tools are mentioning that as their main advantages as most linux systems don't provide those features on their own.
<Noclip><dftxbs3e "let's package snapper then"> nixpkgs already include snapper, maybe that can help you.
<dftxbs3e>Without YaST snapper looks to have no GUI, not very good
<Noclip>Guix doesn't seem to be distro that wants to be used by new users who rely on GUIs.
<dftxbs3e>Timeshift looks like a better time-investment
<Noclip>Timeshift seems to have both GUI and CLI, so yes.
<dftxbs3e>Noclip, DejaVu backup tool looks better for backing up user data
<Noclip>dftxbs3e: I'm not an expert on that topic xD
<Noclip>I partly asked my initial question to find out if there is already another programm or way to automate snapshot creation on btrfs.
<Noclip>But as I said I don't need those tools currently so only package it if you have the time and motivation for it.
<mjlbach>Anyone running mixed guix and nix, and running into issues where the nix packages shadow the packages in scope in `guix environment`?
<mjlbach>guix container works fine ofc
<dftxbs3e>mjlbach, you can use --pure to guix environment
<mjlbach>I did try that, but unfortunately same issue
<mjlbach>Which I thought was weird given my experience with nix shell
<dftxbs3e>mjlbach, I don't use nix so I hope you find out why that is
<mjlbach>No worries! Thanks for the pointer
<mjlbach>It's definitely because the guix environment profile is after my nix system profile on $PATH
<mjlbach>I'm guessing it's because nix modifies path in /etc/zshrc, which places it ahead of the guix environment modification to path
<mjlbach>Ok, fwiw, it's the modification of path in (`export PATH="$HOME/.nix-profile/bin:/nix/var/nix/profiles/default/bin:$PATH"`) removing this seems to work in case anyone else runs into this!
***MidAutumnHotaru0 is now known as MidAutumnHotaru
<apteryx>mjlbach: that doesn't explain why 'guix environment --pure' wasn't enough, unless I'm missing something
<mjlbach>I'm not sure exactly why, but my PATH is `/home/michael/.nix-profile/bin:/nix/var/nix/profiles/default/bin:/home/michael/.guix-profile/bin:/home/michael/.config/guix/current/bin:/bin:/usr/bin:/usr/ucb
<mjlbach>` with `--pure`
<mjlbach>exactly sure *
<mjlbach>I know it's coming from pathmunge in /etc/zshrc (on fedora) populating path with elements from `/etc/profile.d` which includes the nix daemon script that prepends path with the nix entries
<Whyvn>hello, just install guix system. How can I add a repo for binary packages. Specificly guile-emacs
<pkill9>there seems to be a gap in the "market" for a wordpress clone for static sites, i.e. able to just select themes and apply them, edit pages online in one place, all stored as markdown text files
<dftxbs3e>Whyvn, do you have substitute server configured and it's signing key authorized?
<Whyvn>dftxbs3e: its a fresh guix install, havent changed anying configuration. was just trying to get my enviroment setup. so unsure.
<Whyvn>completly new to guix system
<Whyvn> is where it seemed it was pulling the packages from
<dftxbs3e>Whyvn, it seems guile-emacs failed to build on it after time-out, and about the it asks you during install if you want to enable it or not IIRC
<dftxbs3e>Whyvn, guix weather guile-emacs
<dftxbs3e>Whyvn, also
<gregds91>mariadb 10.5.8 fails to build on armhf (after 3+ attempts).
<gregds91>my computer is not running out of space or RAM.
<Whyvn>dftxbs3e: whats guix weather? I had to boot into my debian machine so cannot necessarily check the guix settings atm, I can boot back up to it and install an irc client to relay back here
<Whyvn>trying to gather info here before going back to a packageless install
<dftxbs3e>Whyvn, it checks for substitutes on a particular package
<Whyvn>dftxbs3e: did I miss interupt in that there are no binary packages atm?
<dftxbs3e>Whyvn, yes because either guile-emacs does not build or takes way too long and gets terminated by timeout on CI
<Whyvn>ok thank you, guess I have to take the time to compile
<dftxbs3e>Whyvn, I ran a build myself see if it compiles for me
<Whyvn>is there a resource for configuring grub to add my other OS on different harddrives? Would I just add their drive locations in the config file under the file-systems?
<dftxbs3e>Whyvn, I'm pretty sure that second one wont work, I don't know myself, try the manual
<Whyvn>dftxbs3e: thank for the help, booting back into the guix install to finish configuring
<gregds91>It fails during the 'check' phase.
<dftxbs3e>efraim_, lfam: we must somehow warn for this: - rust-comrak only >=0.9.1 is patched, not older, so we have older packaged and used in GNU Guix (including in recent fractal package by leoprikler)
<lfam>Can you send a bug report?
<lfam>We need to update that package?
<dftxbs3e>lfam, yes but also remove all older versions but some applications depend on older versions and patched one isnt semver compatible
<dftxbs3e>lfam, for bug report I am almost falling to the ground I am aaaa.. tried
<lfam>Maybe there is a patch we can apply to the older version
<dftxbs3e>tired **
<lfam>I'll send the bug report then
<dftxbs3e>lfam, been mechanically iterating over the CVE RSS feed from NVD
<dftxbs3e>lfam, thanks a lot!
<lfam>Do we have a comrak package?
<lfam>I'm not finding it with `git grep`
<dftxbs3e>lfam, sorry then its only added by bug 44492
<lfam>Alright, I'll send a message there
<dftxbs3e>lfam, thanks a lot, that's it for today for me, I encourage checking NVD CVE RSS feeds, been fixing several in GNU Guix looking at them for few hours.
<gregds91>A test 'versioning.partition' is failing.
<newline[m]>I want install kde plasma wihth guix OS is that possible?
<bone-baboon>So I tried again as I wanted to better describe what happend when I booted. So the grey backround and Guix logo are the background image for the Grub menu. After I select the only option in the Grub menu the menu goes away and it is just showing the Grub background image. Again no terminal and it is not responsive to keyboard input.
<bone-baboon>dftxbs3e: You asked about error logs. The boot process has not been consistent anther time I tried booting I was stuck where it said "Welcome to Grub!" with a blinking cursor. However it was not responding to keyboard input. Then an error message appeared about missing something grub related. Was not able to record the error message as it disapeared from the screen.
<bone-baboon>dftxbs3e: I will have to read the Grub documentation to learn how to use the Grub command line.
<bone-baboon>Here is the error message I was talking about. "error: file /gnu/store/hash-grub-hybrid-2.04/share/grub/unicode.pf2 not forund."
<bone-baboon>After that error message it goes to the Grub boot menu. Then it is stuck at a message "Booting 'GNU Guix Intsallation 1.2.0-1.bd96e6d" with a blinking cursor that does not respond to keyboard input.
<marusich>Hello Guix
<vagrantc>bone-baboon: i *think* the error you mentioned is non-fatal
<vagrantc>bone-baboon: but sounds like the kernel never actually successfully loaded or something odd like that
<vagrantc>used to seeing that with u-boot, but not grub :/
<marusich>I'm trying to rewrite the history of the wip-ppc64le branch, but I get an error. Anyone know why?
<marusich>"error: failed to push some refs to ''"
<marusich>It says "remote: error: denying non-fast-forward refs/heads/wip-ppc64le (you should pull first)" but I did do "git fetch". I tried "git push --force-with-lease -n origin wip-ppc64le", then verified it would do the right thing, then ran "git push --force-with-lease origin wip-ppc64le", but I got this error. I tried with --force, too, instead of --force-with-lease, but I got the same error...
<marusich>Do I need to delete it and recreate it?
<marusich>OK, I'll try that.
<marusich>"remote: fatal: bad object 0000000000000000000000000000000000000000" heh
<marusich>It seems to have worked.
<marusich>Despite that strange message
<lfam>I think it's from the remote hook that checks if commits are signed
<marusich>It kind of freaked me out a little
<lfam>It's really weird
<lfam>Terrible user experience
<lfam>$z40 is used by git as the null hash
<marusich>Do you see that when you rewrite your wip branches, too?
<marusich>ok, good, that makes me feel better about it
<lfam>If you check the guix pre-push git hook, you'll see it handles this case specially
<lfam>I don't recall the details of the server-side hook...
<vagrantc>and this is why i've always just merged the wip-pinebook-pro branch :)
<lfam>Perhaps it's not from that hook. It seems like it should handle this too
<lfam>Well anyways, I think it's harmless
<marusich>Do we need to add "Signed-off-by" when cherry picking...?
<marusich>I want to cherry pick e1ea62ceb425447c926a1fdd99095c0ee9a8be07 onto wip-ppc64le so that lle-bout can make commits and it won't cause problems when later we try doing "guix pull".
<marusich>However, if I cherry pick it, it will rewrite the signature, using my key. That's fine, really, but I wonder what others do. I have cherry picked things in the past without adding Signed-off-by, but I didn't think much about it.
<marusich>Sorry, that's the wrong hash, I meant to write 65fb3a3eb9a759e3dba18e78113e80d7e5b353f4
<marusich>I'll just take it as is
***sneek_ is now known as sneek
<marusich>BTW lfam do you know anything about the context for ce5f9afcd1b3f79f254de6a77775f2c2fe6800fe ("gnu: openjpeg: Enable tests")? This re-introduced the same change as commit 49d38b9a44d1045133e993f372394f9d3c626a72, which was later reverted in commit 6af3045f668efe3f1e6eda30e8f7d42148e238b7 by mbakke
<marusich>I ask because the re-introduced test suite is causing trouble on the wip-ppc64le branch. I would like to either disable some of the tests, or disable them all by reverting the commit, but I am not sure which is preferred.
<marusich>Since there appears to be some history here related to openjpeg that you, mbakke, and raghavgururajan might know about, I thought I would ask first.
<marusich>Basically I just want to know: should I disable just the tests that are causing problems for me, or should I revert the entire commit like Marius did?
<marusich>Either way is fine with me.
<lfam>I don't remember the context now
<leoprikler>dftxbs3e, efraim_, lfam: I think we should try applying as a patch to all previous versions
***efraim_ is now known as efraim
<nckx>Good actual morning Guix.
<AllanAdair[m]2>Good actual morning!
***apteryx is now known as Guest66400
***apteryx_ is now known as apteryx
<PotentialUser-7>I'm trying to use guix installed Hexchat and Openssl over TOR to connect to this freenode channel. hexchat seems to expect a cacert pem to be present in a very specific directory that I don't have. ""/gnu/store/9h9153akbd6g2520mcl2d0vgbvyfi5qm-openssl-1.1.1i/share/openssl-1.1.1i/cert.pem", O_RDONLY) = -1 ENOENT (No such file or directory)" -> even
<PotentialUser-7>if I add a cert.pem with the let's encrypt certifificates, they will be lost when I update hexchat and it depends on another openssl version. How do I get out of this bind?
<PotentialUser-7>(correction: the certs directory is there, just not the cert.pem file)
<khassanov>Hey guys. I'm trying to setup `hostapd` on my home server with Guix system. It does not start AP described in `/etc/config.scm`. How can I check logs to see why it fails? I see log files in `/var/log` but wondering is there is a better way to read service log. Thanks!
<PotentialUser-7>FYI, I solved my problem by installing nss-certs and opening a new shell. Now I get other problems.
<mdevos>khassanov: Guix System doesn't have a special logging facility ... I believe GNOME has a system log viewer you might like, but I can't find the guix package (maybe it isn't packaged?)
<khassanov>Ok, thanks mdevos ! I see `/var/log/messages` and `/var/log/${daemon_name}.log` have logs necessary.
<cbaines>Hmm, make check-system TESTS=childhurd fails for me :(
<cage_>mdevos, i sent a bug report this morning about compilation error on debian testing we talked yesterday
<mdevos>cage_: I found it in my inbox.
<mdevos>I'll look into it, but currently I'm working on something else and it's progressing nicely
<mdevos>That said, does this bug block you from using guix?
<cage_>mdevos, no problem i am not in a hurry! :)
<cage_>i think i'll compile the code with that form commented out and try to build the packages i would like to submit to guix
<cage_>if they build i think i can submit the patch
***chrislck1 is now known as chrislck
***terpri_ is now known as terpri
<mdevos>cage_: keep in mind Debian doesn't always have the latest guix, so you may need to "./pre-inst-env guix pull" to get a recent guix (make sure --localstatedir is set correctly) that includes your patch
***koleesch_ is now known as koleesch
<cage_>mdevos, thanks for your suggestions!
<narispo>marusich: hello!
<pretzel>Hi. Is there a (semi-)official guix.git mirror? (If yes, where?) Thanks.
<johannes`>Hi, I've never used IRC before, but I've search the archives of both
<johannes`>the IRC and guix-help and can't find anything.
<johannes`>I'm trying to use the graphical GUIX system installer but can't get
<johannes`>past the partitioning screen. I've already done the partitioning so
<johannes`>I'm trying to proceed via the so-called manual partitioning screen.
<johannes`>All I have to do is set one of the partitions are root "/" but there
<johannes`>seems to be no way to do it. I can delete a partition and add it
<johannes`>again but how does one set the mount point?
<cbaines>I'm unsure the graphical installer supports picking existing partitions, rather than creating them
<cbaines>I'm not that familiar with it though, so I could be wrong
<raghavgururajan>Hello Guix!
<nckx>pretzel: No official mirrors, <> is administered by people very close to the project.
<nckx>Hi raghavgururajan.
<johannes`>That may be the case. I'm happy to try anything. Have tried deleting and adding partitions.
<raghavgururajan>marusich: Let me take a look. Guix's git ( is not loading.
<johannes`>When one adds a new partition one does get a list of items relating to that partition but I don't know how one edits that list from within the partitioner and there seems to be no guidance at all.
<johannes`>The list includes the mount point but it is empty. I just need to edit it somehow.
*raghavgururajan /me just realized that there is a mirror at
<nckx>raghavgururajan: URL?
<cage_>is down?
<raghavgururajan>nckx: [Guix]: [Guix/Guix/]:
<raghavgururajan>* nckx: [ALL]: [MAIN]:
<Rovanion>Native-inputs is where I put python requirements for the check/test phase right?
<raghavgururajan>nckx: You are co-owner. xD
<nckx>Pum pum pum. ♪
<raghavgururajan>marusich: It seems the tests are disabled ( What's the issue then?
***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | ⚠️ is down, some commands won't work | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | FOSDEM: | This channel's logged: https://logs'
***nckx changes topic to 'GNU Guix | ⚠️ is down, some commands won't work | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
***ChanServ sets mode: -o nckx
<narispo>raghavgururajan: we work with core-updates on wip-ppc64le
*raghavgururajan wonders why 49d38b9a44d1045133e993f372394f9d3c626a72 was reverted as 6af3045f668efe3f1e6eda30e8f7d42148e238b7
<raghavgururajan>narispo: I see. openjpeg fails to build there?
***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | ⚠️ is down, some commands won't work (try --url= | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
<narispo>raghavgururajan: yes, because some tests fail, not sure if legitimate failures or not, but we disable conformance tests in addition to nonregression and it works now but still a bit worrying
<raghavgururajan>narispo: Oh. How there are tests with `#:tests? #f`?
<raghavgururajan>May be the revert (6af3045f668efe3f1e6eda30e8f7d42148e238b7) is not in core-updates?
<nckx>raghavgururajan: Because the tests fail on an architecture (I think it was arm, but I'm not very certain).
<nckx>Such things should be explained in the commit message.
<raghavgururajan>narispo, marusich: Okay, I am clear now, Tests are enabled in core-updates, which fails on wip-ppc64le. Yeah, so to your initial question, it is better to disable failing tests instead of disabling all.
<raghavgururajan>nckx: Gotcha!
<nckx>(I mean, in addition to ppc64, but I'm not sure.)
<raghavgururajan>I think it is better to do 'if condition for arch' in tests, instead of disabling tests globally.
<raghavgururajan>Damn, I spent long time on enabling tests on openjpeg with openjpeg-data, and testing it on x86_64. But its okay I guess.
<raghavgururajan>LoL. I can now see performance hit on Disroot's Git.
*raghavgururajan notifies Disroot's admins to expect extra load today
<raghavgururajan>Wow, Gittea recovers very fast.
<nomad-cr32>any one has guix package receipe to build grub from git head?
<PotentialUser-7>I've got most icons working in evince, except for the one at the top left (side pane), which still looks broken. I've tried installing various *-icon-theme packages, and running fc-cache -f each time (even fc-cache -r), to no avail. Has someone managed to get the side pane icon to work?
<raghavgururajan>nomad-cr32: You mean latest grub? `guix pull && guix build grub`
<nckx>I think they mean from GRUB's git head, not ours.
<raghavgururajan>Ah I see.
<raghavgururajan>> various *-icon-theme packages
<raghavgururajan>PotentialUser-7: Did that include hicolor-icon-theme and adwaita-icon-theme?
<nomad-cr32>raghavgururajan, yes
<PotentialUser-7>raghavgururajan: yup, and just to be sure and just ran again "guix install hicolor-icon-theme adwaita-icon-theme && fc-cache -rv", and opened another terminal, same problem
<nomad-cr32>i have one prepared, but there seems to be an issue gnulib
<raghavgururajan>nomad-cr32: You can use the HEAD's commit in the pack-def.
<raghavgururajan>nomad-cr32: Could you paste the log?
<raghavgururajan>PotentialUser-7: That's odd. Btw, are you using Guix on foreign distro or Guix System?
<PotentialUser-7>raghavgururajan: Guix on foreign distro (Ubuntu).
<raghavgururajan>PotentialUser-7: Also, when running evince from terminal, do you see any warnings/errors related to missing icons?
<PotentialUser-7>raghavgururajan: I see this: Could not load a pixbuf from /org/gtk/libgtk/theme/Adwaita/assets/bullet-symbolic.svg.
<PotentialUser-7>(evince:19116): Gtk-WARNING **: 17:11:25.042: Could not load a pixbuf from /org/gtk/libgtk/theme/Adwaita/assets/bullet-symbolic.svg.
<PotentialUser-7>This may indicate that pixbuf loaders or the mime database could not be found.
<PotentialUser-7>But I don't know what to make of it
<raghavgururajan>PotentialUser-7: I see. I not sure which evince you are running. Evince installed by Guix or Ubuntu. Can you run `guix environment --pure --ad-hoc evince hicolor-icon-theme adwaita-icon-theme -- evince`
<PotentialUser-7>$ echo "$(readlink -e $(which evince)) $(evince --version)"
<PotentialUser-7>gives: /gnu/store/kjd09h5imyfhhc0nfgbsj5n1w1md4ljb-evince-3.36.5/bin/evince GNOME Document Viewer 3.36.5
<nomad-cr32>raghavgururajan, here is the gist, not very clean, but works
<PotentialUser-7>raghavgururajan: running in the ad-hoc environment gives a broken icon as well, but the icon looks differently broken
<PotentialUser-7>raghavgururajan: maybe the icon is simply not in any icon theme
<raghavgururajan>PotentialUser-7: Ah. Looks like a bug. Could you report it?
*raghavgururajan is afk
<nckx>PotentialUser-7: Check whether there's a previous report first, this bug is ooold.
<nomad-cr32>raghavgururajan, here is the failing logs
<PotentialUser-7>raghavgururajan:, correct? what standard information will be required? I could paste the IRC log to start with, and add a screenshot
<raghavgururajan>PotentialUser-7, nckx: Probable fix is replacing ("gdk-pixbuf" ,gdk-pixbuf) with ("gdk-pixbuf" ,gdk-pixbuf+svg). I may have fixed this already in wip-desktop.
*raghavgururajan is now really afk
<nckx>I looked for a previous report without luck, so no idea. I do know it's been around forever. If it's fixed on w-d that would be grand.
<nckx>This is frustrating. I know I took screenshots of broken Evince icons for the same reason.
<PotentialUser-7>reported the issue
<PotentialUser-7>by email, with links to to
<raghavgururajan>nomad-cr32: I am not sure of the errors. May be someone else can help you.
<raghavgururajan>PotentialUser-7: What is the bug number?
<mzcore_>What is the current status have zfs in guix? would I have issues on guix+zfs doing fresh install?
<nckx>There's no support for Guix ‘on’ ZFS, if that's what you mean. But using ZFS file systems ‘from’ a Guix System should work.
<PotentialUser-7>raghavgururajan: I sent the report to, did not get any acknowledgement or automatically assigned issue number yet, looking at the issue tracker
<nckx>PotentialUser-7: Don't wait too hard, it's probably stuck in moderation (hey, that's me!) and being partially down makes that rather hard to do.
<nckx>I've got no mail so far.
<PotentialUser-7>Message-ID: <>
<PotentialUser-7>Subject: Bug report: broken side pane icon in evince
<PotentialUser-7>I'm not blocked or anything
<nckx>Have you sent mail to our bug list before?
<PotentialUser-7>yes, successfully
<nckx>Then you may or may not be blocked or anything.
<PotentialUser-7>last time i did *not* send a screenshot attachment, if that makes any difference
<nckx>The Guix AI® modera-tron™ is a complex black box whose operation is a mystery to even the gods.
<nckx>Nah, it wouldn't.
<PotentialUser-7>I'm all in on guix, so I'm not going anywhere. If I don't get a reply, I'll re-submit, I have it in my outbox, so it's not lost.
<nckx>It shouldn't be lost, just wandering.
<nckx>For now, just assume that services resuming will unblock your mail.
<nomad-cr32>raghavgururajan, thanks
<raghavgururajan>PotentialUser-7, Fixed ^
<raghavgururajan>nckx: Would you be able to push that fix?
*nckx .oO to where?
<raghavgururajan>Never mind. Hahaha
<mzcore_>nckx: yes, I mean system installation like curses or even manual
<nckx>Guix System can't currently boot from ZFS.
<nckx>There might be experimental patches out there but I haven't heard of any.
<mzcore_>got it, thanks
<mdevos>there are some ZFS patches on the mailing list / issue tracker. idk if they include ZFS on root file system, though
<nckx>mdevos: No, not yet.
<nckx>/home yes, / no.
<mdevos>speaking of filesystems, can the childhurd be booted from some network file system? I want to the childhurd to have its root directory as some directory of the parentlinux
<nckx>milkman[bot]: Really? Again?
<nckx>We talked about this.
<nckx>The HTML, it is not hard.
<mdevos>(to avoid having to mess with partitions and to avoid having to choose a partition size)
<samplet>Is Savannah down?
<mdevos>samplet: it is down for me
<nckx>See topic.
<samplet>Ah okay. Thanks for the quick response.
<nckx>Most of FSF/GNU is down.
<nckx>Including mail, Web, … :-/
***nckx changes topic to 'GNU Guix | ⚠️ GNU/FSF networks down, some commands won't work; try --url= | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
<samplet>Any speculation as to what’s happening?
<nckx>Speculations abound, but I haven't found any authoritative answers 😉
<raghavgururajan>samplet: jxself in #fsf, says: "Fear not. There are surely alarms going off for the sysadmins. "To the GNUmobile!" they exclaim as they slide down the pole and speed away to the GNU Data Center buried deep under a Stallman-shaped mountain."
<nckx><jx self> Fear not. There are surely alarms going off for the sysadmins. "To the GNUmobile!" they exclaim as they slide down the pole and speed away to the GNU Data Center buried deep under a Stallman-shaped mountain.
<nckx>So that's one.
<nckx>Super original quote, raghavgururajan.
<samplet>Thanks! (I should probably ask these sorts o’ questions on #fsf in the first place....)
<raghavgururajan>nckx: Wow our brains are in sync
<nckx>I usually add <> to the topic but it's all crickets too.
<bandali>or #savannah
<nckx>I lurk in all of them, haven't seen any explanation?
<samplet>nckx: That is where I looked, too.
<nckx>‘GNU/FSF seem to be experiencing (partial?) network outage’, which is, admittedly, true.
<bandali>if this is like the last similar outage, it's probably due to a faulty network switch in the fsf infra
<bandali>still waiting for word from one of the fsf sysadmins though
<nckx>These things have built-in Sunday-sensors dontchaknow.
<nckx>raghavgururajan: Are we causing noticable load for disroot? I'd be surprised, but we can use a different mirror if they prefer.
<kisaja[m]>I think they got ready in time
<raghavgururajan>nckx: I can myself see the difference, but gitea is recovering very fast. I already notified admins and they are okay with it. Nothing bad. :-)
<nckx>Our own (stale) mirror is embarrasingly 502ey.
<raghavgururajan>nckx: I don't see performance difference anymore. So they must have allocated few more RPUs.
<nckx>Raspberry Pie Units?
<kisaja[m]>Is there a list of mirrors. Maybe someone doesn't trust disroot
<raghavgururajan>(RPU = RAM + CPU)
<nckx>Never heard of it, thanks.
<bone-baboon>01:26 <marusich> ok, good, that makes me feel better about it
<raghavgururajan>*I meant to say RAU (taken from
<raghavgururajan>nckx ^
<nckx>kisaja[m]: Each commit is signed, and the commits currently served check out. ‘guix pull’ checks GPG signatures. Even if went evil, we should notice. That said, I'm not happy about it either. We *should* have our own mirror; it's broke.
<bone-baboon>Please disregard my last two messages they were sent in error.
<nckx>No problem.
*nckx holds breath.
***ChanServ sets mode: +b fsf*!*@*
***fsf-gnu was kicked by ChanServ (User is banned from this channel)
<nckx>What a shock that turned out to be porn.
<kisaja[m]>Where to expect news post, they survived
<nckx>Can it be true?
<nckx>Holy shazam.
***nckx changes topic to 'GNU Guix | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
*raghavgururajan reports to #riseup
<nckx> FSF/GNU is back. \o/
<nckx>raghavgururajan: Cool!
<nckx>(That image is disgusting and I do apologise.)
<dftxbs3e>well it's just porn not illegal content or anything don't think it's worth reporting to riseup nor do I think it qualifies for deletion really just inappropriate here
<nckx>PotentialUser-7: ...and your mail arrives.
<raghavgururajan>Hey Guix Folks! The following is greetings from Muppeth of Disroot:
<raghavgururajan>‎2021-02-28 | 12:26:24 PM ‎Muppeth‎: nice to give helpng hand though
<raghavgururajan>‎2021-02-28 | 12:26:29 PM ‎Muppeth‎: send my greetings to guix crew raghavgururajan
<mdevos>nckx: is down for me?
<nckx>dftxbs3e: Legality don't really enter into it. Their service is being abused by skids to annoy people; some like to know.
<nckx>mdevos: Ah, but it's less down (it's sending gibberish) and all ‘we’ currently care about is Savannah.
<nckx>And mail.
<raghavgururajan>> nckx‎: (That image is disgusting and I do apologise.)
<raghavgururajan>You don't have to. That spammer is a disgusting person.
<raghavgururajan>> ‎dftxbs3e‎: well it's just porn not illegal content or anything don't think it's worth reporting to riseup nor do I think it qualifies for deletion really just inappropriate here
<raghavgururajan>If the service is used to spam, then its an abuse of the service. This is not the first time in #guix. Same person, same type of spam.
<nckx>raghavgururajan: OK, bug #46831 isn't the exact same issue that I have. Here, icons are rendered, but with errors: <>. Hence my suprise that a +svg variant would fix it, since that already looks like a (perhaps build-time) SVG rendering to me.
<nckx>I hadn't even noticed the missing icon next to that.
<raghavgururajan>> ‎PotentialUser-7‎: (evince:19116): Gtk-WARNING **: 17:11:25.042: Could not load a pixbuf from /org/gtk/libgtk/theme/Adwaita/assets/bullet-symbolic.svg.
<raghavgururajan>nckx: My patch targeted this.
<Whyvn>hello, i recall there was a way to view how an installed packages schm configuration. what was the command to do this?
<dftxbs3e>Whyvn, guix package --export-manifest ?
<raghavgururajan>It seems evince used gdk-pixbuf to load icons. But without +svg variant, the engine is not loading svg files.
<nckx>raghavgururajan: Right. This one must be rendered elsewhen then.
<Whyvn>dftxbs3e: maybe? that command didnt seem to work. I want to create a package that is very similar to one that is offered the the guix repo and was wanting to view the one in the guix repo as a reference for creating
<dftxbs3e>Whyvn, oh, then maybe try EDITOR=nano guix edit <pkg>
<raghavgururajan>PotentialUser-7, nckx: I now see "Gtk-WARNING **: 12:44:27.010: Could not find the icon 'view-sidebar-symbolic-ltr'. The 'hicolor' theme was not found either, perhaps you need to install it."
<Whyvn>dftxbs3e: thats it! thanks
<nckx>...why nano...
<nckx>raghavgururajan: Did it fix the icon?
<dftxbs3e>nckx, user-friendly terminal editor (on screen keyboard control documentation) with open file at line support
<raghavgururajan>nckx: I looked at the wrong icon. No, the patch didn't fix the one we were looking for (view-sidebar-symbolic-ltr). But the patch fixed the loading of another icon called bullet-symbolic.svg.
<nckx>I'm not that familiar with GTK to know what's what, but progress is progress.
<nckx>dftxbs3e: nano is already the default.
<nckx>What's the best video editor for people who don't video-edit, and also in Guix?
<nckx>I need to make a 110 GiB video less 110 GiB.
<nckx>So something stable would be preferred.
<ngz>If reducing size is all you need, vlc, among others, can do this, IIRC.
<dftxbs3e>nckx, try kdenlive, if re-encode to reduce size with quality try ffmpeg
<nckx>Thanks. I just need to cut, ideally without quality loss. Encoding is optional, I can do that on the CLI too.
<dftxbs3e>nckx, pitivi also
<ngz>nckx: e.g.,
<dftxbs3e>nckx, with pitivi looks simple click the cut button on the right after sliding your video file onto the timeline and positioning the cursor where you want to cut, it splits the video import in multiple separate elements you can then delete and reposition, once done just render
<simonsouth>Looks like is up again.
<simonsouth>...minus the extra 'g'.
<Whyvn>if a package is on the gnu package website, but not in the cli guix search, what could cause this?
<nckx>Thanks for all your answers ( laptop crashed -_-)'. Must admit it hadn't occurred to me to use VLC as a video editor.
<nckx>Whyvn: Which package is it?
<nckx>Will try pitivi first.
<nckx>nomad-cr32: I'm guessing you got git #grub to work? 😉
<muradm>@nckx, just built for now :) i don't know if it works ) will reboot and see
*nckx does not have a Freenode-wide auto-alert set for ‘guix’, it just seems that way...
<Whyvn>nckx: lemonbar-xft
<nckx>Whyvn: ‘guix search lemonbar-xft’ returns 1 package here, so I misunderstand your question. Which command does what, and what should it do?
<nckx>It's a relatively ‘new’ package (Dec 2020); make sure your guix is up to date (guix describe).
<Whyvn>nckx: ok, ill have to update guix, thanks for confirming.
<Whyvn>couldnt update guix atm because compiling guile-emacs for over 10hrs
<Whyvn>dont want to have to restart it
<muradm>@nckx, this builds
<muradm>only efi support, didn't bother with others
<muradm>at least it shows what to do with sources and what should be patched
<muradm>if one will decide to make "grub-git" and friends packages for guix :)
<muradm>it is too late here, i will be testing this in the morning, if my laptop will boot with luks2 encrypted boot
<Rovanion>I'm trying to package a python library and its tests fail due to the source path being /tmp/guix...drv-0/django-ajax-selects-2.1.0 instead of /tmp/guix...drv-0/src, do I just add a build phases before and after check that renames the source dir?
<Rovanion>Or is there a better way?
<nckx>Whyvn: 10h? 😳 Wow.
<nckx>muradm: It might be worth adding ‘grub-next’ just for the LUKS2 support.
<nckx>Thank you.
<nckx>Rovanion: Can you patch the source to replace the hard-coded ‘src’ with (basename (getcwd)) instead? Or, even better but potentially holey and full of rabbits: find out why it hard-codes the expected build directory name like that to begin with.
<efraim>Want worse? I'm working on powerpc support for guile 3.0.3, just spent 74 hours compiling 3.0.2-73, and that was with skipping tests.
<nckx>I just hope Whyvn's build is actually building.
<Whyvn>yeah unsure if its frozen or not.. top says processes are running and guix processes shows it. if it doesnt finish by the time i get home i am going to close it and just cut it as a loss
<nckx>Rovanion: I think renaming the build directory is a bit beyond ‘supported’, even if it happens to work now.
<Rovanion>nckx: Its not a hardcoded 'src` but rathe that python fails to load modules with dots in their paths:
<nckx>Dat nasty.
<Rovanion> is down btw :/
<nckx>Whyvn: Sometimes looking at strace -p PID and seeing something obviously loopy can remove vain hope.
<nckx>< is down btw :/> FFS.
<nckx>The Sunday detectors are doing good work today.
<dftxbs3e>nckx, been the case since yesterday at least
<lfam>We can type our pastes into the google search bar and then share them via auto-complete
<nckx>dftxbs3e: Mkay. I'll change the one in the topic again. Whose politics do we agree with again?
<nckx>I usually use gnome's but their default expiry is only 30min.
<lfam>I was gonna suggest GNOME
<lfam>I don't think we are going to find many "friendly" paste sites
***nckx changes topic to 'GNU Guix | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged:'
<dftxbs3e>nckx, I know some encrypted pastebins because they wont store plain text content
<nckx>Done. Not an endorsement, etc. (then neither is Debian's).
<Rovanion>I'm not even sure how the folder-name ends up there, its not part of the tarball.
<nckx>dftxbs3e: ‘Encrypted’ stuff tends to require JS by design, so they are straight out for #guix. Not for political reasons, but because I insist on ‘being able to curl shit to git’ as a feature.
<dftxbs3e>nckx, yes..
*nckx AFK; if Serious Freedom Issues are found with p.g.o, feel free to amend.
<dftxbs3e>nckx, but not storing plain text is avoiding legal struggle for hosters
<lfam>Yeah, there are good reasons that these sites use JS
<dftxbs3e>I believe there would be interest in developing standard tooling for encrypted paste/download
<lfam>But, I'd rather use it to fingerprint visitors than to encrypt the content
<lfam>Encrypted anonymous filesharing is a risky proposition
<joshuaBPMan50>Hey #guix people!
<raghavgururajan>Gajim's PasteBin can do paste via pipe and get paste via curl.
<raghavgururajan>nckx: ^
<leoprikler>gotta love those /api links
<Rovanion>Any tips on how to change the directory that's used to store the source being built?
<lfam>IIUC, you can use the TMPDIR environment variable
<leoprikler>or what exactly are you trying to do?
<lfam>That moves the entire build directory to $TMPDIR. You have to set it in the context of guix-daemon
<lfam>leoprikler's answer is also correct. It depends on what question you're asking :)
<leoprikler>cmake-build-system also has #:out-of-source? #f for cases where that's a problem
<Rovanion>Right, thet problem is that Python doesn't like folders that contains dots: so when I try to run the tests for library that I'm packaging Python complains that it can't import the library because it can't find it.
<Rovanion>So I need to change the directory that the source code is extracted to so that Python becomes happy.
<Rovanion>Though I have no idea why it tries to adress the source from its parent directory to begin with.
<leoprikler>that's weird, but you can try (mkdir "dummy") and then do some stuff there
<leoprikler>it seems to be just for testing as well, so perhaps that's also an indicator for something
<Rovanion>I did stupid thing of adding `(mkdir "dummy")(chdir "dummy")` in a phase before 'unpack but that of course just ended with the source folder being inside "dummy". $TMPDIR contains the path where the source will then be extracted, not the name of the source directory :/
<lfam>It's surprising, I don't recall seeing this problem before with the dots before
<Rovanion>This is my package definition:
<Rovanion>Building it should error out.
<leoprikler>I don't think you should do this before unpack
<leoprikler>the thing that errors out is check, so you should do your chdir before check and maybe replace the check phase with something more appropriate
<Rovanion>I'm not sure where I could chdir for it to make a difference. I need the source dir to not contain a dot.
*nckx back. Isn't this just a silly Python bug?
<leoprikler>i'm pretty sure it's a bug in the test suite only
<leoprikler>honestly, you could just as well just fix the suite
<nckx>I just read ‘thet problem is that Python doesn't like folders that contains dots’.
<nckx>raghavgururajan: Thanks, I'll try to remember that next time we have to play the fun musical pastebin game.
<leoprikler>Python doesn't like folders containing dots *for module names*.
<leoprikler>You don't name your Guix modules (foo #{1.0 } dirty/stuff) either.
<nckx>Challenge accepted?
<mdevos>Hi Guix! I'm half-seriously contemplating including emoji and/or various other Unicode symbols in some source files. Is this known to cause reproducibility issues?
<lfam>I don't see why they would
<leoprikler>Guix source is UTF-8 afaik.
<lfam>We already have a bunch of λ
<civodul>mdevos: if these are meant to go to stdout/stderr, there should be a fallback for non-utf-8 locales
<civodul>(although i hope we can consider non-utf8 locales a thing of the past)
*lfam pushes their first news entry
***Noisytoot is now known as [[
***[[ is now known as Noisytoot
<civodul>lfam: wo0t!
<nckx>Good news or bad news?
<lfam>Kernel update
<lfam>So, I suppose it depends :)
<nckx>Weird, doesn't show here.
<leoprikler>Improved Wine performance!
<nckx>Of course it doesn't show here if I'm still working around Savannah being down, smh.
<nckx>Shit, now I need to translate ‘kernel’ too.
<nckx>‘de kernel’. Ah, Dutch, how do I love thee.
<lfam>That's what I head, leoprikler!
<lfam>I mean, that's what I heard
<lfam>We compared the English and French words for kernel yesterday
<vagrantc>stone pit!
<lfam>Funny word. It's mainly a metaphor in English
<lfam>The only time you'd use it literally is talking about corn
<vagrantc>occasionally other grains ...
<lfam>Yeah, but that's pretty obscure
<lfam>But "kernel of truth" is a common idiom
<lfam>And, corn kernels aren't even on the inside usually
<vagrantc>english took the idiom "consistancy is the hobgoblin of mediocre minds" a bit too far
<Rovanion>leoprikler: Thank you! Had tried to find any reference to the parent directory in the librarys codebase and was starting to look at code of django-admin. Should have just RTFM I guess :D
<nckx>lfam: I think that well illustrates the meaning of core or nucleus (Dutch: kern); it's not a seed.
<vagrantc>wait a minute, this is #guix, surely someone will have to go into the nuances of the translation to esperanto!
<vagrantc>or is it treated as one of those "foreign" words?
<Rovanion>Do you also have "pudelns kärna" in your languages, as in "the real problem/matter/issue/meanning"?
<lfam>The heart of the issue
<leoprikler>Des Pudels Kern in German, I think it's a Goethe reference.
<leoprikler>Those Germans, not even using Jojo references smh.
<Rovanion>Yes, apparently from Goethes Faust.
<Tirifto[m]>vagrantc: If I understand correctly that the talk is about ‘kernel’ as an operating system component, then the usual word in Esperanto would be ‘kerno’ indeed, generally meaning ‘core’ or ‘pip’.
<Tirifto[m]>(Not to be confused with the Python package manager.)
<bandali>incoming intermittent outages in the gnu/fsf infra
<ATuin>hi, seems that `guix environment -C` is broken, operation not allowed when mounting, anymore having this issue?
<civodul>i had to look up "corn kernel"; in French, "corn kernel" and "corn" are the same word
<lfam>I never thought of that
<lfam>It's funny how these words are borrowed but the meanings become incompatible
<lfam>It also explains why I've noticed people from other countries using "corns" to refer to, like, a bowl of kernels
<cbaines>I still can't explain the mathematical use of "kernel"... although I have some recallection of its connection to aspects of machine learning
<lfam>As Marilynne Robinson said, "words are poor things"
<ATuin>in swedish korn means barley :D
*rekado always imagined des Pudels Kern to be a little squishy
<jonsger>lfam: nice one, pretty "Promiment" ^^
<luis-felipe>Hi, what would be the correct way to write (list bind "utils") as a specification for specification->package?
<luis-felipe>I tried (specification->package "bind:utils") but it fails.
<lfam>The manual has an example for specifications->manifest, if that helps
<lfam>In the section Invoking guix package
<ss2>hi, I'm trying to create a partition with a higher inode count for guix. Which parameters do you preferebly use?
<ss2>I've fiddling with mkfs.ext4 -i or -I, but then the inode counts are either very high, or the freshly created partition is already a bit too full.
<efraim>I normally use btrfs
<mdevos>About the question about Unicode symbols in source files: I meant Unicode symbols in the source file *name*.
<ajarara>I don't see a dhclient pid file (or it running), and attempting to provide it over desktop-services errs with 'service networking provided more than once'
<ajarara>Is networking a composite service? I don't see a declaration of it in gnu/services/networking.scm
<ATuin>what's the best way of testing an old commit of guix, building from the repo or using guix pull --commit?
<ATuin>seems `environment -C` is broken and i'm trying to know which commit broke it
<dftxbs3e>ATuin, time-machine ?
<ATuin>ahh, never used that
<ss2>efraim: I still don't dare touch it after once having a weird data-los.
<ATuin>dftxbs3e: thanks, looks exactly what i'm looking for
<dftxbs3e>ATuin, guix time-machine --commit=<blah> -- environment --ad-hoc hello -- hello
<lfam>ss2: So, you want more inodes, but not too many?
*mdevos Zzz
<ATuin>ehh that command is wonderful :D
<ss2>yes, I think I'm settling in for -i 2048 at mkfs
<luis-felipe>lfam: Using the guile specification that includes the debug output in specifications->manifest also fails :(
<luis-felipe>I don't understand what I'm doing wrong...
<luis-felipe>For example, in a guix repl:
<luis-felipe>(use-modules (gnu packages))
<luis-felipe>(specification->package "guile@3.0.2")
<luis-felipe>Returns a package.
<luis-felipe>(specification->package "guile@3.0.2:debug")
<luis-felipe>guix repl: error: guile: paquete no encontrado para versión 3.0.2:debug
<luis-felipe>"No package found for version 3.0.2:debug"
<rimarko>Hi everyone, I'm a new user of Guix, could you help me understand how I can make the mercurial python module available to the python interpreter? I've installed the Mercurial package of course, and python2 as well... I'm a bit lost.
<civodul>luis-felipe: try 'specification->package+output'
<lfam>ajarara: Are you saying that, when using %desktop-services, there's no dhcp daemon?
<luis-felipe>civodul: Ah, ok, thanks.
<civodul>luis-felipe: that one returns two values
<luis-felipe>civodul: Can I use that in the list of packages of the operating system definition?
<luis-felipe>I guess not?
<nckx>lfam: Why's Wine performance first and Btrfs performance last?
<civodul>luis-felipe: no because it returns two values, so the second would be truncated
<lfam>No reason nckx
<lfam>I copied the highlights from the kernel newbies article
<luis-felipe>Ok, thanks.
*nckx wonders if it's beyond the translator's purview to combine them.
*civodul .oO( still no Dutch translation of the web site )
<nckx>I think I'm at 16%.
<nckx>It's not... fun.
<civodul>oh you started? neat :-)
<nckx>Oh, I thought that's what you were referring to.
<civodul>no, that was meant as a subtle suggestion :-)
<civodul>i didn't know you had already started
<ATuin> <- seems this is the commit that broke the -C in environment
<nckx>It's not for tomorrow, but what would be the steps to ‘activate’ it? I've never used Weblate, nor Haunt, nor, well, most of this stuff.
<lfam>I don't think it's very important to include those highlights in the Guix news entry, rather than just alerting users to the update. But then I figured that I might as well add hype
<ATuin>not a big one :)
<civodul>ATuin: what/how did it break? could you report a bug?
<ATuin>yeah, what's the procedure to do it
<civodul>nckx: last time we discussed it with roptat, we thought any translation above some threshold, like one third, could be put online
<ATuin>`guix environment -C bash` returns `Operation not permitted`
<civodul>ATuin: email with all the details
<ATuin>ok, i will do it tomorrow
<dftxbs3e>civodul, I saw someone else having the same issue here on IRC earlier
<ATuin>i tied the previous commit and it worke fine
<civodul>ATuin: for the record, was about fixing an "Operation not permitted" issue
<ATuin>funny :D
<civodul>well, can't wait to see bug reports :-)
<civodul>with mount flags and all
<cbaines>The childhurd system test has problems due to building the hurd package, /gnu/store/swwaszcprnpph49bn74l06rmi7qwssb5-hurd-0.9-1.91a5167.drv
<cbaines>I think this is a recent breakage, but I haven't tracked it down yet, has anyone got any suspicions?
<nckx>civodul: Noice! I was working in a rather linear fashion but will now prioritise strings with that in mind.
<cbaines>I'll try to get back to working on system test stuff in the Guix Data Service at some point, which will hopefully make tracking down breakages like this a little easier
<civodul>cbaines: argh, no idea
<raghavgururajan>nckx: Just to check, do you get my DMs?
<ajarara>lfam: Yeah, I have %desktop-services in a modify-services form like this:
<cbaines>confusingly, for a revision where the it's possible to at least run the tests (and one where I'm running working childhurds, the hurd package also fails to build), so I think it might be more complicated than a package breaking
<cbaines>that's which is from around two weeks ago
<lfam>ajarara: I'm surprised, but it seems like you have to add the service. I would have expected it to be included in %desktop-services
<lfam>Or at least for network manager to handle it
<ajarara>I can try inlining the values literally.
<civodul>cbaines: looks like this is the first broken one:
<civodul>so that'd be in the 5c7874adb00c834e55e58a9b964ebc5fd1bb872c..fc03a1a3d793f4497f5fac560a724a63db51196c range
<cbaines>nothing in that range jumps out...
<nckx>raghavgururajan: DMs have been got. But as you know, I hardly check them. I've done so now. Thanks!
<civodul>cbaines: maybe it's just non-deterministic?
<roptat>nckx, the process is manual, but we take care of it, you don't have to worry
<lfam>ajarara: If I understand correctly, network manager handles DHCP
<lfam>And that is part of %desktop-services
<raghavgururajan>nckx: Cool!
<lfam>So, it should be working...
<ajarara>lfam: sudo herd status shows networking as a service
<roptat>it's all exlained in the guix-artworks repository, adding a language is a bit tricky, and updating one is simply updating the po file once it's added
<lfam>ajarara: That's not really a service
<lfam>Different services can provide "networking", which is what you see there
<lfam>So, DHCP isn't working, right? It's not just the lack of a dhclient that you are worried about?
<nckx>roptat: That's good, but I'd like to worry out of curiosity, since I'd like to fix the (vaguely) related <> but the current system looks strange & complex to me.
<ajarara>lfam: Yeah, running dhclient gets me to a working network setup
<cbaines>civodul, I'm unsure, I'd first question if that's actually the range where the breakage occurred
<nckx>I mean, I'm sure there's a good reason for it, that's not a complaint.
<lfam>ajarara: You might have to go into network manager GUI and set things up
<luis-felipe>It seems the documentation about "specification->package" led me astray.
<luis-felipe>It says "(list bind "utils")"
<ajarara>lfam: nm-applet works as well, though I had to put in the WPA creds, and now I've got two ip addresses under `ip addr`
<cbaines>there's been a bunch of builds for the hurd package, and they've all failed with that same error as far as I've seen
<luis-felipe>But it seems the package for bind is called isc-bind to avoid a conflict with a built-in Guile procedure.
<lfam>I'd guess the Guix desktop assumes use of network manager, rather than dhclient
<luis-felipe>So it should be (list isc-bind "utils")?
<lfam>The package is named bind, but the variable that defines it is called isc-bind
<civodul>cbaines: "guix time-machine --commit=3c548c3e0e -- build hurd --target=i586-pc-gnu" passes
<lfam>So, it depends on if you are listing packages or variables, luis-felipe
<cbaines>civodul, that's good, but I'm not sure that derivation is what's causing the system test to fail
<luis-felipe>lfam: I was trying to add bind:utils to the list of packages of my operating-system definition.
<ajarara>another idea is to filter out network-manager-service-type and sub in dhclient.
<lfam>The sky is the limit with service composition ajarara
<lfam>At a certain point, it's easier to start with %base-services and build things up, than to remove things from %desktop-services
<cbaines>civodul, the Guix Data Service doesn't seem to know about the derivation I see fail /gnu/store/swwaszcprnpph49bn74l06rmi7qwssb5-hurd-0.9-1.91a5167.drv which maybe says something... I'm not sure where it comes from?
<cbaines>civodul, it doesn't match guix build --no-grafts -d hurd
<civodul>cbaines: not sure where it comes from then
<dftxbs3e>cbaines, hello! does your last commit changing configuration of GNU Guix mean my configuration will break on my GNU Guix System VM with unattended-upgrade enabled?
<cbaines>dftxbs3e, yeah, unfortunately, it's a breaking change to the configuration record.
<dftxbs3e>cbaines, can't do depreciation etc?
<luis-felipe>I mean, in Using the Configuration System, Globally-Visible Packages, it seems it should be (list isc-bind "utils") instead of (list bind "utils").
<cbaines>dftxbs3e, I'm not sure it's worth it in this case. I think taking a look at your agents might be good anyway, as I haven't seen much activity recently
<cbaines>dftxbs3e, I could be wrong though, the instrumentation still needs work
<lfam>luis-felipe: Are you able to send a patch?
<dftxbs3e>cbaines, my agents are up on running but there's bottleneck on substitution I think, things are sitting mostly idle due to bottlenecks
<cbaines>dftxbs3e, I think there are bottlenecks, but I think there's still a few cases where substitution indefinately hangs, which isn't great when it happens. I'm going to make another pass over the code once is done
<dftxbs3e>cbaines, don't your builders also sit idle most of the time?
<cbaines>dftxbs3e, the fosshost ones are active currently. Not as fully utilised as I'd like, but they haven't got completely stuck
<cbaines>dftxbs3e, if you restart the service, it might spring in to life again
<luis-felipe>lfam: I think so. I will try tomorrow.
<dftxbs3e>cbaines, doing the configuration update now, mine are also active but really slow, especially PowerPC 64-bits ones, I think it's guile being slow there
<raghavgururajan>Latest commits from lfam and cbaines doesn't appear on CI.
<cbaines>raghavgururajan, Cuirass polls the repository, so it's not always going to be up to date
<raghavgururajan>Ah Cool!
***ece3 is now known as ece