IRC channel logs


back to list of logs

<nckx>Although HTTPS works for me again now.
<nckx>Oh, no, it doesn't, git was stuck. All is well and broken 👍
<rekado_>we should configure and make it proxy to
<dGtlc3>Hi folks. I made myconf.scm(copied all from /etc/config.scm) to change some keyboard stuff, but after `guix system reconfigure myconf.scm` I get the same configs. What am I doing wrong?
<vagrantc>trying to update a machine i haven't updated for a couple weeks, but it sits for ... hours at: substitute: updating substitutes from ...
<vagrantc>figured it was just a network glitch, but it's happening again...
<str1ngs>vagrantc: try removing /var/guix/substitute/cache/ . I'm not sure of a better way to invalidate this
<str1ngs>also try curl maybe it's a network issue?
<bandali>great talk, dongcarl :) the only thing i felt missing and would’ve loved to hear you mention was the freedom aspect perhaps
<vagrantc>str1ngs: thanks, will try
<vagrantc>does core-updates get rebased, or is it generally fast-forward?
<nckx>dGtlc3: Nothing, according to your description. The file name doesn't matter.
<nckx>kmicu: I missed your success. Glad to hear it.
<kmicu>[jokin’] Not being able to fetch Guix commits (for ~5 minutes) was a very dramatic experience. Please FSF Infra Staff—never again!
<kmicu>BTW does guix pull depend on being online or are fallbacks set up?
<dGtlc3>Damn, I was hoping it was me, that way I can fix it
<vagrantc>kmicu: you can explicitly pull from another repository with --url
<kmicu>dGtlc3: could you change something else than keyboard stuff in myconf.scm and reconfigure again to confirm that config file works at all?
<kmicu>vagrantc: thank you.
<dGtlc3>ye, I changed timezone to test that, still same
<dGtlc3>is there a specific location myconf.scm needs to be?
<kmicu>Timezone can not work b/c of ntp not syncing when the difference is more than x minutes. Better to change something less confusing xD
<kmicu>dGtlc3: did you restart your system/shell after reconfiguring?
<dGtlc3>kk, gonna change few more
<kmicu>(I also assume you’ve executed guix system reconfigure mycomf.scm as root/sudo)
<rekahsoft>Hi all, it seems that is not working
<vagrantc>rekahsoft: seems to be the case, yes...
<dGtlc3>:D ofc
*kmicu 😴
<Marlin[m]>hi guix
<Marlin[m]>I got a bunch of stuff on lua and minetest modding
<Marlin[m]>making some snippets for minetest on my emacs as well
<dGtlc3>same :|
<dGtlc3>lol nice timing :D
<dGtlc3>by same I mean configs still not changing
<Marlin[m]>oh :P
<vagrantc>is anyone else's guix-daemon running with a large number of --chroot-directory /gnu/store/... --chroot-directory /gnu/store.. arguments?
<vagrantc>i saw this on another machine earlier, and now it started happening on this latest machine too
<vagrantc>is that expected?
<vagrantc>seems like my proxy is not caching and rejecting the ssl cert....
<vagrantc>which is related to my earlier issue
<Marlin[m]>hi guix
<Marlin1113>./configure; sudo make bootstrap
<Marlin1113>luarocks mostly needs this as arguments
<Marlin1113>how do i pass that for it on the packaging stuff?
<khtkgnulinux>I'm a new guix user, run guix on CentOS , run the 'guix pull' command , but return error: 'unexpected HTTP status code: 502'.
<khtkgnulinux>I try install GuixSD on lenovo ideapad 100s (chromebook) but the initrd is bokern, How can I manual to build initrd.
<recj>hwhat do you do when a package assumes standards fhs ?
<khtkgnulinux>run it on lxc or docker
<dmarinoj>vagrantc: The jetson tx2 is in the u-boot mainline! I'm going to try making a bootloader in Guix and chainloading from the eMMC as suggested here
<vagrantc>sneek: tell dmarinoj jetson probably requires binary blobs to sign it so probably can't make it into guix
<sneek>dmarinoj, vagrantc says: jetson probably requires binary blobs to sign it so probably can't make it into guix
<vagrantc>sneek: tell dmarinoj later jetson probably requires binary blobs to sign it so probably can't make it into guix
<sneek>dmarinoj, vagrantc says: later jetson probably requires binary blobs to sign it so probably can't make it into guix
*vagrantc shrugs
<bavier`>vagrantc: "later tell ..."
<vagrantc>sneek: later tell dmarinoj jetson tx2 probably requires binary blobs for boot firmware (e.g. u-boot) so probably can't be integrated into guix
<vagrantc>sneek: botsnack
<leungbk>`gtags --skip-unreadable --skip-symlink` <- is this the right way to generate gtags for the scheme files in the guix repo? i appear to only get hits for the C files.
<bavier`>leungbk: idk that gtags works well for scheme source. otoh, I've heard geiser is a great tool for finding your way around scheme projects. I've personally not taken the time to really learn it.
<leungbk>the only thing i don't like about geiser for the guix repo is the fact that i have to use geiser-eval-buffer on every buffer to make the definition-hopping work, which takes some time.
<leungbk>well, every buffer in which i want to call the definition-jumping
<leungbk>thanks though
<bavier`>leungbk: hmm, ok
<arbi>Is it recommended to use guix to install rust packages?
<arbi>how can I package rust crates?
<kmicu>dGtlc3: just a check: are you on a working Guix System or during an installation process? If the latter then maybe you did not execute ‘herd start cow-store /mnt’.
<kmicu>sneek: later tell khtkgnulinux it’s possible that Guix System installer didn’t generate a proper base-initrd-modules entry in the config. So you need to add it manualy.
<sneek>Will do.
<kmicu>arbi: Guix has cargo-build-system and you could use it to define a Rust package
<mbakke>Good morning Guix!
<kmicu>Mornin’ 🤠
<mbakke>The Haskell changes on 'staging' looks good, I would like to merge them before pushing a more intrusive patch set to that branch.
<mbakke>rekado: WDYT?
<rekado>mbakke: staging has just been merged a few days ago, no?
<rekado>I don’t mind merging more often, but the branch needs to be fully built.
<kmicu>Did someone in the past force push to staging branch?
<mbakke>rekado: The two new commits on the branch are already fully built.
<rekado>merging is fine then
<dGtlc3>gm :)
<dGtlc3>kmicu: fresh 1.01 install
<g_bor[m]>Hello guix!
<g_bor[m]>I have a slight problem.
<g_bor[m]>It seems that I would like to introduce a circular dependency into the module graph, which breaks stuff. However, the package I am defining belongs to that module, I am quite sure about that. What would be the best way to fix this?
<g_bor[m]>What I tried was to use select in the use module, to select only the package I need, but that does not work.
<mbakke>g_bor: What is the module cycle? Probably one of the packages will have to be moved.
<g_bor[m]>mbakke: it is benchmark and logging. I believe that a package named benchmark belongs to benchmark. I pushed it earlier today, and works fine. The other is a c++ logging library, I can imagine that it could be moved from logging. Wdyt?
<mbakke>g_bor: Including (gnu packages benchmark) from logging.scm should be fine, no?
<g_bor[m]>Should be, but it does not work.
<g_bor[m]>mbakke: another option would be to disable benchmarking on the logging library
<g_bor[m]>maybe I have something stale here, I am trying a make...
<g_bor[m]>the error I get is a bunch of failed to load module, unbound variable in module-lookup.
<g_bor[m]>As soon as I remove (gnu packages benchmark) it works fine. This looks like a dependency cycle to me...
***furrymcg1e is now known as furrymcgee
<kmicu>dGtlc3: so to clarify is Guix System already installed? Could you paste (on a paste service) ‘guix describe’ and ‘guix pull’ output?
<polezaivsani>Hey guix! Can somebody help me - i'm trying to install docker with guix on a foreign distro, but after having installed docker package, i don't see any dependencies (e.g. containerd) pulled into my profile. What am i missing?
<pkill9>polezaivsani: you might need the docker-cli package
<polezaivsani>pkill9: yeah, i'll try it
<dGtlc3>kmicu: Ye.
<polezaivsani>pkill9: as i thought, the docker-cli gives you just the cli tool. And i still need to run the dockerd, which fails with the dependencies missing.
<pkill9>polezaivsani: when i go into a guix environment with docker, it download containerd as a dependency, but when i run dockerd it doesn't find containerd in the $PATH. I think you'll just have to install containerd as well
<pkill9>i think a patch for the package definition for docker wrapping dockerd with a $PATH pointing to containerd would be accepted
<polezaivsani>pkill9: is this the right way to go? i.e. doesn't having containerd as dep for docker mean that the dependencies will be installed too? i would be glad to figure it out and come up with a patch
<kmicu>Thank you dGtlc3: what about the output of ‘sudo guix system reconfigure /path/to/myconfig.scm’ (Please use /path/to/myconfig.scm not ~/myconfig.csm just be sure ;)
<polezaivsani>pkill9: with containerd installed manually, dockerd still fails for some other reason - wonder if i'd need to do so all the other packages mentioned in it's depencies list
<dGtlc3>I get this: failed to activate service org.freedesktop.Accounts requested by `:1.9` while booting pre gdm, maybe there is connection
<pkill9>polezaivsani: what's the new error?
<kmicu>dGtlc3: now the fun part: remove a random paren from myconfig.scm and execute reconfigure again.
<polezaivsani>pkill9: it's about devicemapper, but it's prob due to kernel config - i tried version from my native package manager and it shows the same. i'll tell more once i finish tweaking the kernel config
<Marlin[m]>hi guix
*Marlin[m] sent a long message: < >
<civodul>Marlin[m]: you probably need to customize the 'configure' phase to pass appropriate flags
<Marlin[m]>how do i remove that flag?
<Marlin[m]>gonna test it without it
<civodul>to remove it, you need to replace the 'configure' phase
<civodul>there are several packages that do this
<dGtlc3>kmicu: I get an error, depending on which paren: missing closing paren, invalid field specifier, source expr failed to match any pattern
<anon43881>hello guys. can somebody tell me how I can get xorg working on guix. I installed xorg-server, xinit and drivers like xf86-video/xf86-input, but startx does not find X.
<pkill9>anon43881: the startx command from xinit is broken, I think there may be a service that provides a working startx command, I know there is a guide in an email somewhere that says how to get a working startx command, it's quite old now though
<pkill9>i dunno why it hasn't been fixed so it atleast works, even if it doesn't included a global system configuration as it's just a static package
<polezaivsani>pkill9: with a proper kernel, dockerd seem to be working with containerd installed manually.
<nckx>pkill9: (service special-files-service-type `(("/bin/startx" ,(xorg-start-command)))
<polezaivsani>pkill9: i'll investigate this issue further in a while. thanks for helping out!
<pkill9>you're welcome :)
<Marlin[m]>hey guix
<Marlin[m]>how do i use qmake for a package?
<pkill9>Marlin[m]: typically replace the configure phase with a phase that invokes qmake
<pkill9>there's some examples in other packages
<Marlin[m]>i'm trying to package smplayer
<kmicu>dGtlc3: ok, so that config file is read correctly and everything should work. 🤷
<kmicu>dGtlc3: so I assume that if you add ‘hello’ to packages ‘(packages (cons hello %base-packages))’, then execute ‘sudo guix system reconfigure myconfig.scm’, (then to be sure execute ‘hash guix’) then ‘hello’ should be in your path. So the real issue is only with the keyboard layout, yes?
<emsyr>Hello. Anyone using enlightenmet?
<emsyr>Hello. Anyone using enlightenmet? I'd like to ask about connman.
*kmicu is using connman but not Enlightenment.
<Marlin[m]>god i'm terrible at packaging :P
<nckx>emsyr: Don't ask to ask; just ask ☺
*nckx uses neither 🤷
<emsyr>kmicu: connman does not work when logged in as typical user. Here is log when I type connmanctl :
<Marlin[m]>i'm trying to package these two
<Marlin[m]>any ideas about what's the issue?
<emsyr>Here is the message for connman service in enlightenment : . It is related to dbus service.
<emsyr>kmicu: nckx: connman works fine when logged as root.
<kmicu>emsyr: let me check how connman behaves on XFCE..
<dGtlc3>kmicu: Your assumption is correct. So ye, only keyboard layout is the issue. I thought /etc/config.scm reflects current configuration
<nckx>dGtlc3: It does, after you've reconfigured your system to match it.
<kmicu>dGtlc3: so let’s debug that keyboard issue now.
<kmicu>dGtlc3: so ‘(keyboard-layout "us" "intl"#:options '("caps:swapescape"))’ doesn’t work? Could you remind us the original issue please?
<dGtlc3>well that's basically the issue
<kmicu>emsyr: connmanctl works with a regular user on XFCE. So Enlightenment requires some patches.
<emsyr>kmicu: Enlightenment seems little bit unstable. I have crashes during use. I like it though.
<emsyr>kmicu: Is there is anything I can do to help debugging?
<saslibre>echo "hello"
<nckx>dGtlc3: You've probably done so above, but do you have a link to your current system.scm?
<nckx>kmicu: Thanks.
<saslibre>I've installed guix, but i have a graphics problem >.<
<nckx>Why are there two .scm files, are they identical, and which one is being used to test?
<kmicu>emsyr: is caps not swapped during boot or in X? You did not set keyboard layout for Xorg like on
<kmicu>nckx: myconfig.scm
<kmicu>(It’s in one file to avoid pasting many links.)
<nckx>Well, kmicu already pointed out the obvious, so…
<nckx>kmicu: s/emsyr/dGtlc3/ ?
<emsyr>kmicu: Keyboard issue is not mine :)
<kmicu>dGtlc3: so I think you need to replace the last line with ‘(services (cons (set-xorg-configuration (xorg-configuration (keyboard-layout keyboard-layout))) %desktop-services)’ and that should work.
<nckx>dGtlc3: Is your keyboard layout incorrect in GRUB or the Linux VT?
<kmicu>(Assuming Xorg layout is not changed*)
<nckx>Hm, I'm going to stop ‘helping’ because I've obviously missed something. Suddenly it's about Xorg…
<nckx>& this is why we do one thing and do it well, nckx.
<dGtlc3>Ahh my bad
<kmicu>nckx: yeah, that issue is from yesterday.
<dGtlc3>I read the first example not the second one
<emsyr>saslibre: ?
<kmicu>dGtlc3: no problem.
<dGtlc3>I'm gonna add to bootloader and xorg now and test it
<saslibre>I'm having problems with X11, i've just installed guix
<kmicu>emsyr: feel welcome to send a bug report cuz I cannot find anything about that;include=subject%3Aconnman
<emsyr>kmicu: Ok, thanks.
<g_bor[m]>saslibre: hello, what kind of problem do you have?
<saslibre>My xorg isn't starting, and my keyboard bloqnum led starts blinking
<saslibre>I don't know if this helps, in /var/log/messages: gdm-session-worker: too many open files
<saslibre>gdm_session_set_environment_variable value !=NULL
<saslibre>Also i had to add pata_acpi on initrd-modules, since guix system init asked
<kmicu>emsyr: I see that Entl. explicilty extends dbus-root-service-type and XFCE not. So that’s probably the issue but Entl. maintainers should know better what to do with that.
<kmicu>saslibre: you have the same issue as quiliro 🤔😺
<saslibre>I'm with quiliro right now!, we're trying to install guixsd on the same machine that gives us the issue
<saslibre>We have days with the issue >.<, so if you need more information just ask :D
<emsyr>kmicu: I'll fill a bug report asap.
<kmicu>Thank you emsyr.
<kmicu>saslibre: could you check whether you have /var/lib/gdm/.cache/ or ~/.cache/gdm ?
<emsyr>kmicu: I thank you for the help :)
<kmicu>saslibre: for now I assume maybe there’s an issue with gdm cache cuz quiliro mentioned that the disk doesn’t have a fresh /home.
<kmicu>saslibre: you could also use a pastebin service (e.g. and share /var/log/messages and /var/log/Xorg* files.
<kmicu>(and /var/log/gdm* if exists)
<saslibre>There's no cache of gdm in any of those two directories, also, in /var/lib/gdm is empty
<kmicu>(or /var/log/slim assuming it exists)
*kmicu went to plant some beans.
<saslibre>/var/log/gdm does exist
<saslibre>ON /var/log/gdm/greeter.log has several errors, one of them is: Failed to load module "openchrome" (module does not exit, 0)
<saslibre>I assume that the module openchrome since, it doesn't load, we don't have a graphics user interface
<saslibre>The VGA controller that we have does use openchrome, but openchrome isn't loaded
<saslibre>What is the latest version of guix? I want to test i have the latest version
<dGtlc3>I just installed latest image and updated, mine is 1.0.1-3.4
<saslibre>What does guix --version tell you?
<dGtlc3>guix (GNU Guix) a60dbad3bc8211cdec2f623306fc2e02f1cc17a1
<Marlin[m]>i'm packaging lutris
<Marlin[m]>getting permission errors
<nckx>Marlin[m]: Paste 'em!
<saslibre>dGtlc3: Thanks!
<Marlin[m]>one sec
<nckx>saslibre: Have you added xf86-video-openchrome to the (xorg-configuration)'s (modules) list?
<nckx>It's not included by default.
<saslibre>nckx: no
<saslibre>it was not necessary on 0.16
<nckx>You're getting several test errors, you might have to disable that last one (or not; I didn't look into it). The /homeless-shelter one means the test tries to write to $HOME, which is set to that non-existent directory by default.
<nckx>Marlin[m]: ☝
<nckx>Marlin[m]: That can be easily ‘fixed’ by (setenv "HOME" "/tmp").
<nckx>Plenty of examples in gnu/packages/*.scm. grep away.
<Marlin[m]>thanks :)))
<nckx>Marlin[m]: If you're lucky that will fix all but 1 failure ☺
<saslibre>It seems that i don't have the package... Do I install it?
<saslibre>Also, the xf86-video-openchrome wasn't on config.sch
<saslibre>It doesn't install automatically
<nckx>saslibre: Are you talking to me?
<nckx><nckx> [Add] xf86-video-openchrome to the (xorg-configuration)'s (modules) list? It's not included by default.
***drrrty is now known as drrty
<saslibre>nckx: what is the exact synthax
<saslibre>nckx: or where do i add the xf86-video-openchrome?
<nckx>saslibre: No idea. I only know that xf86-video-openchrome was never part of %default-xorg-modules. It was never removed. If you didn't need it in 0.16, I wonder why X suddenly complains about it now.
<Marlin[m]>i'm not lucky :P
<Marlin[m]>now this issue
<nckx>saslibre: I think the (set-xorg-configuration …) is what you're looking for, but I don't use that myself [yet] so I'm not familiar with it [yet].
<nckx>s/ is/ service is/
<nckx>Marlin[m]: That's not right. It's failing before the test suite now… Could you paste the package?
<saslibre>nckx: I'm working on it right know, thanks
<nckx>saslibre: I can't give you exact syntax, but it could look something like this: (services (cons (set-xorg-configuration (xorg-configuration (modules (list xf86-video-openchrome xf86-input-libinput …)))))
<nckx>Of course that needs to be adapted to your configuration (cons is probably not appropriate) and is 100% untested.
<nckx>And the whole openchrome error might be a red herring to begin with.
<saslibre>nckx: Thanks for all the help. I'm working on it!
<saslibre>I'll keep you updated
<nckx>saslibre: I hope it's a productive outing.
<nckx>Marlin[m]: That makes more sense than the first ☺
<nckx>(Marlin[m]: Feel free to nckx: me, I'm doing other things™ and not keeping much watch here.)
<nckx>Marlin[m]: Why I muck about with that: note that the version's ‘v’ needs to be moved to the URL, and that you should be using git-fetch here instead of GitHub's auto-generated /archive/ tarballs, which can change (this has actually happened!)…
<Marlin[m]>oh ok
<Marlin[m]>Any ideas?
<Marlin[m]>can i still get the stable release with git-fetch tho?
<nckx>Marlin[m]: Eh, the package runs its tests… in the build phase (in fact that seems to be the only thing it does there) so it wasn't what I thought.
*nckx continues poking.
<nckx>Marlin[m]: Everything offered as a tarball is offered over git, and much more.
<nckx>(commit (string-append "v" version)) is the magic word, since COMMIT may also be a tag, et voilà, a release ☺
<nckx>(Again, this pattern is very common in gnu/packages/*scm, you can look for examples by grepping for the snippet above.
<nckx>Marlin[m]: Also add a (file-name (git-file-name name version))
<nckx>I just ran ‘guix build lutris --source’ and it returned /gnu/store/9ffrx85hxnrpbjwify7dw9r00d5m29b6-v0.5.2.tar.gz
<nckx>Not a good name for a tarball ☺
<nckx>(That's purely for humans, though, nothing to do with your build failure.)
<nckx>lutris/util/ ‘Return a list of available libraries, as returned by `ldconfig -p`.’
<nckx>debian/changelog: * Use ldconfig to determine library paths
<nckx>Marlin[m]: (add-after 'unpack ' (lambda _ (substitute* "lutris/util/" (("ldconfig") "true")) #t))
<nckx>Now I get:
<nckx>ERROR: Failure: ImportError (cannot import name GnomeDesktop, introspection typelib not found)
<nckx>ERROR: Failure: ValueError (Namespace WebKit2 not available)
<nckx>That's one very entitled test suite.
<nckx>Marlin[m]: I'm going to hand it back to you now. If there's no way to make those tests work without those 2 ridiculously heavy dependencies (*and* they're not actually needed to run this package), just disable them. That's unreasonable.
*nckx notices… you might want to change the name of that phase. Whups.
<Marlin[m]>i'm gonna try it later
<Marlin[m]>If you wanna change it and see if it works, many thanks :P
<saslibre>nckx: I'll chat from the pc that has the problem, with quiliro username
<nckx>Marlin[m]: Have you tried building the package with tests disabled (#:tests? #f), just to know whether it works as-is?
<nckx>quiliro: All right, I'm going AFK in a few minutes though.
<nckx>Marlin[m]: Change it? It's just a name. Just not a name that should make it into Guix.
<quiliro>nckx: hi!
<nckx>We could name half of our phases unfuck-foo otherwise.
<Marlin[m]>hey guix
<Marlin[m]>what's the right method for installing nix?
<Marlin[m]>guix install nix works, but trying to install stuff with nix-env returns "no permissions to make /nix"
<nckx>Marlin[m]: Nix is like Guix: it needs a daemon to perform superuser duties. Did you set up the daemon?
<nckx>nix-daemon-service-type, or so.
<nckx>Whatever it's called.
*nckx → AFK as promised.
<Marlin[m]>ooh ok thanks
<saslibre>nckx: We did the config, but no good
<saslibre>The same problem, but we added xf86-video-openchrome to the config.shm
<saslibre>nckx: We keep having no video
<quiliro>gdm: Could not start command '/gnu/store/0jp8hwdj6lsjbqrqp46w3b347jigmpx5-gdm-3.28.2/libexec/gdm-session-worker': Too many open files
<quiliro>that after:
<quiliro>gdm: gdm_session_set_environment_variable: assertion 'value != NULL' failed
<Marlin[m]>How does guix import nix work? It's bugging here
<quiliro>what is the replacement for build-utils
<quiliro>sorry build-essential
<str1ngs>quiliro: gcc-toolchain
<quiliro>str1ngs: thanks
<str1ngs>no problem
*nckx .oO Guix is so much better at naming things…
<nckx>saslibre: I was afraid of that since you said that it worked fine on 0.16.
<quiliro>> This is my messages file: , i have no X11, and the NUMLOCK LED flashes
<thomassgn>um, trying to start cow-store service, but herd says could not be found...
<quiliro>This is my config file:
<quiliro>thomassgn: what is the exact command you are using?
<thomassgn>is the cow-store service not available on running/installed systems? Can I reenable it somehow?
<thomassgn>sudo herd start cow-store /raid
<thomassgn>which yields 'herd: service 'cow-store' could not be found'
<quiliro>ls /raid?
<Marlin[m]>i'm getting this thing when i try to import a package from nixpkgs
<quiliro>does this work? ls /raid
<thomassgn>ls raid gives the expected folders as output. /raid is mounted, just copied the last backups before moving the system
<thomassgn>I mean ls /raid
<nckx>thomassgn: No, cow-store is a work/hackaround for the limited RAM-drive in the installer.
<quiliro>could this be the problem: `lib.nixpkgsVersion` ?
<nckx>Not a general purpose thing. Although you could always re-use it yourself.
<str1ngs>I think think cow-store should be called barn. just saying :P
<nckx>str1ngs: Stop cowshedding.
<quiliro>str1ngs: :-D
<rekado>thomassgn: it is not available on installed systems.
<str1ngs>oh I see what you did there! :P
*rekado didn’t see nckx’s correct response
*str1ngs paints his chow shed white with black splotches
<thomassgn>nckx, rekado : is cow-store necessary for 'guix system init ./config.scm /raid' to work?
<nckx>str1ngs: Chow? Cows are friends, not food ☹
<rekado>cow = copy on write
<thomassgn>hahaha, barn >)
<nckx>thomassgn: Not if your /gnu/store has the free space required.
<thomassgn>awesome. thanks. :) nckx
<rekado>isn’t the point to arrange for writes to /gnu/store to be written to disk?
<quiliro>gdm: Could not start command
<quiliro> '/gnu/store/0jp8hwdj6lsjbqrqp46w3b347jigmpx5-gdm-3.28.2/libexec/gdm-session-worker':
<quiliro> Too many open files
<rekado>quiliro: I guess you have too many files open.
<g_bor[m]>hello guix!
<quiliro>and gdm: gdm_session_set_environment_variable: assertion 'value != NULL'
<quiliro> failed
<rekado>there’s a limit of file descriptors that may be open at the same time.
<nckx>quiliro: You can remove the set-xorg-config again if it doesn't make a difference.
<str1ngs>I think that too many file issues is gdm spawning to many times
<quiliro>it is the same error message as without it
<nckx>rekado: Not to the target's /gnu/store, if that was @ me.
<str1ngs>the issue is probably lack of graphics
<nckx>str1ngs: quiliro definitely lacks graphics.
<str1ngs>I also thing how the gdm service handles this is not so graceful
<quiliro>str1ngs: It could not load the video driver, in this case: xf86-video-openchrome
<str1ngs>maybe it can be limited by max restarts or something. I don't know how shepherd handles this
<str1ngs>or retries?
<rekado>speaking of GDM: after reconfiguring I cannot log in anymore… Had to remove /var/lib/gdm/* again, and now gnome-shell doesn’t start…
<quiliro>How can i load the xf86-video-openchrome module?
<nckx>quiliro: Small correction: we don't know if the Guix package is even needed or relevant here, the original message was:
<nckx> /var/log/gdm/greeter.log has several errors, one of them is: Failed to load module "openchrome" (module does not exit, 0)
<str1ngs>quiliro: it's not a module. its a driver
<str1ngs>X11 driver to be specific
<nckx>str1ngs: Our xorg-configuration calls them modules.
<g_bor[m]>I tired to investigate the cycle that would be produced by using benchmark in logging. The problem seems to be that benchmark uses maths, which in turn uses logging. Do you have an idea what would be a better module for a c++ logging library, so that I can move it to break this loop?
<nckx>str1ngs: That's what they mean.
<quiliro>How should i install the driver?
<nckx>I suggested earlier to try (modules (list xf86-foo)).
<str1ngs>quiliro: you can also use the drivers field in xorg-configuration
<quiliro>nckx: I did, check my config:
<nckx>quiliro: You did. Do you still get *that* error? Failed to load module (← str1ngs 😛) "openchrome"?
<nckx>quiliro: I know.
<str1ngs>hmm if the module does not load it will not use the driver
<Marlin[m]>any ideas about this?
<str1ngs>never mind I confused driver with modules
<nckx>Marlin[m]: Nah, haven't used Nix in ages. Is that all you get? That's basically empty output. Or is everything without Warning an error? In any case, we seem to be speaking some very old Nix to nixpkgs :-/
*nckx thought the Nix integration was brand-spanking new, but guesses that doesn't include the importer.
<Marlin[m]>i just cloned the git repo from nixpkgs really
<quiliro>>here's my greeter.log :ç1LSx
<nckx>Marlin[m]: I'm the worst person to ask about the importer. I use it very seldom and then only for CPAN.
*nckx doesn't trust them robots 😒
<Marlin[m]>hey, how do i use a local repository again?
<str1ngs>use a channel with file://path/to/repo for the url
<str1ngs>err file:///path/to/repo
<kmicu>quiliro: let’s do something else. Could you test ?
<Marlin[m]>hmm, then it should be wroking
<Marlin[m]>do i just pull it afterwards?
<str1ngs>it will only see commit changes. not the work tree
<str1ngs>yes you need to pull
<quiliro>kmicu: What do these changes do?
<nckx>quiliro: Use SLIM (\o/) instead of GDM.
<nckx>Since GDM is both buggy and very hard to debug, and SLIM is old but tends to work.
*nckx biased.
<quiliro`>nckx: remove variable not assigned
<quiliro`>nckx: have you forgotten `(use-modules (rnrs lists))'
<quiliro`>thosse are the messages on reconfigure
<kmicu>It lookes like openchrome is not loaded. And that’s the real issue. I wanted to switch to not-GDM to exclude gdm .cache issues or countless ‘Jun 15 10:46:35 localhost gdm: Child process 3575 was already dead.’ already reported on the ml.
<nckx>quiliro`: That was kmicu's helpful file, not mine. And wait, there's a gotcha with remove: there are two, but I forget which one is the bad one.
<nckx>kmicu: Not loaded how?
<nckx>It's not in the list of default modules, no.
<nckx>Or is it really not found by Xorg even when added to the list?
<kmicu>quiliro`: I forget to add imports for ‘remove'. Could you try this
<nckx>quiliro`: You definitely don't want (rnrs lists)'s remove.
<nckx>…was going to say use SRFI-1's ☺
<kmicu>nckx: it’s not loaded by Xorg. There is no LoadModule: "openchrome" hence it breaks.
<nckx>kmicu: So what does your configuration do to address that?
<nckx>I wonder if (drivers "openchrome") would help here.
<kmicu>As already stated that config is for Slim to exclude GDM cache bugs and other GDM bugs reported on ML cuz quiliro` doesn’t have a fresh /home.
<nckx>OK, so nothing, gotcha.
*kmicu pasted that test config before reading greeter log.
*kmicu checks whether openchrome DDX supports KMS.
*nckx hasn't thought about openchrome since… 2010? Good point.
<kmicu>quiliro`: the output of ‘dmesg | grep -e agp -e drm’ would be appriciated.
<kmicu>(Nvm its VIA P4M890)
*kmicu LOL’s at an article about VIA ‘If your machine freeze at startup (GDM) or after login (SLiM),…’
<bricewge>Where can I find an example to send the result of a mcron job by email?
<kmicu>quiliro`: there is a lot of things to test. E.g. blacklisting viafb.
<bricewge>The doc hint that's possible but I only found a unresponed question about it in the mailing list.
<civodul>rekado: bah, GDM broken again?
*kmicu check whether the recent kernel removed DRM module for Openchrome. 🤦
<kmicu>Is it possible to use a kernel substitute (to avoid compiling) from 0.16 in 1.0.1?
<rubic88>I've got i3lock installed and running, but it won't accept my login (and I have to kill the process remotely). Do I need to configure screen-locker-service?
<thomassgn>rubic88: I think that will do the trick
<thomassgn>for an example see:
<rubic88>thomassgn: trying now, thanks
<kmicu>It looks like there’s no kernel Linux-libre LTS too ;(
<nckx>kmicu: ? We have all of them.
<kmicu>quiliro`: to clarify. I think something changed in recent kernel Linux regarding Openchrome and that’s why it works on older Guix System (with older kernel) but not on new one where it ‘[drm] Failed to open DRM device’.
<nckx>4.19.50, 4.9.181, 4.4.181. The best of Greg.
<kmicu>nckx: with substitutes? No recompilation needed?
<nckx>quiliro`: You can try (kernel linux-4.19) in your operating-system to test an older (supported) kernel.
<quiliro`>with the last config file i get: service 'xorg-server' more than once
<nckx>kmicu: That's a different question. Why would that matter?
<quiliro`>dmesg | grep -e agp -e drm
<quiliro`>[ 0.702382] Linux agpgart interface v0.103
<quiliro`>[ 0.702757] agpgart: Detected VIA P4M890 chipset
<quiliro`>[ 0.708765] agpgart-via 0000:00:00.0: AGP aperture is 128M @ 0xd8000000
*nckx just recompiled 4.19 on an ARM box; it's not that slow,
<kmicu>I don’t want to force someone to compile a kernel on an old hardware just to test my ‘maybe’.
<nckx>kmicu: You're not forcing anyone ☺ It's a very good hypothesis and the best one they've got.
<kmicu>quiliro`: on a working setup you should also have those lines too:
<kmicu> [drm] Initialized drm 1.1.0 20060810
<kmicu> [drm] Initialized via 2.11.1 20070202 on minor 0
<kmicu> [drm] Used old pci detect: framebuffer loaded
<xavierm02_>rubic88: I had that problem at some point. I had installed i3lock as a user (guix install). Installing it for the whole system (putting it in config.scm and guix system reconfigure) fixed it
<kmicu>I understand that but NixOS builds different kernels and we can test them in 1min with ‘boot.kernelPackages = pkgs.linuxPackages_latest;’ and I don’t like when NixOS gives better experience 😹
<rubic88>xavierm02_: If thomassgn's suggestion fails, I'll try your approach. Thanks.
<quiliro`>so i what config should i use now to test?
<kmicu>I forgot that set-xorg-configuration implicitly uses gdm-service.
*kmicu generally doesn’t like implicit arguments.
<quiliro`>guix system: error: se proporciona el servicio 'xorg-server' más de una vez
<quiliro`>that error before including (kernel 4.19)
<quiliro`>if i add (kernel 4.19) i will get that same error
<nckx>kmicu: How is NixOS's experience better? Downloading our 4.19 kernel should be as fast as theirs, once substitutes are built.
<kmicu>NixOS has alternative kernels ready out-of-the-box.
<nckx>quiliro`: It would be helpful to not translate output when pasting it here (even though we can figure out that particular one).
<nckx>kmicu: How don't we?
<kmicu>quiliro`: that’s an issue in my config. Give me a sec.
<nckx>Do you mean that they have more kernels packaged than our mainline + all LTSes?
<nckx>We just don't have the markpower for that.
<kmicu>nckx: can a Guix System user set ‘(kernel Linux-libre-some-LTS)’, reconfigure and have in in 1 minute?
<kmicu>(I don’t know that’s why I’m asking.)
<nckx>If substitutes are available, yes. Same goes for NixOS (although I undestand that they don't pull master directly, so they ‘always’ have substitutes available, or I've been misinformed.)
<nckx>kmicu: It's not like the LTS kernels are blacklisted on the build farm or anything.
*kmicu was (unknowingly) on core-updates-next branch and could not find set-xorg-configuration. 🤦
*rubic88 No luck with i3lock yet, will try xavierm02_'s suggestion to include i3lock in config.scm.
<nckx>kmicu: *whispers* Come to the dark side; we have worktrees.
<kmicu>nckx: so quiliro` should (after I fix xorg error) set ‘(kernel linux-libre-4.19)’?
<nckx>Let me double-check to make sure that's exactly syntactically correct ;-)
<nckx>quiliro`, kmicu: Yes. Import (gnu packages linux) if you haven't already and add (kernel linux-libre-4.19) to the top level of your operating-system.
<nckx>We also have 4.14 and 4.9 if you want to go older.
<kmicu>Anyone care to check whether evaluates?
<nckx>I can confirm that 4.19 downloads in 7 seconds, not those tedious minutes that NixOS apparently takes :o)
<nckx>kmicu: guix system: error: service 'xorg-server' provided more than once
<arkhan>I'm try to install sddm as login manager but not work
<kmicu>nckx: I assume no line numbers from Guile? 😿
<nckx>That was literally the only line from ‘guix system build foo’ :-/
<kmicu>I pasted twice too 🤦🤦
<kmicu>Thank you nckx: did you evaluate only first part? Now fixed
<nckx>kmicu: I passed the whole thing to guix system build.
<nckx>Not that I mind at all, but you could test it yourself with the same command.
<kmicu>Can I do it with sole guix?
<nckx>kmicu: Seems to eval.
<kmicu>Don’t yell at me but I’m on Debian now.
<nckx>kmicu: Not ~ System? Sure. I'm almost certain it supports --dry-run.
<kmicu>quiliro`: try please.
*nckx just uses ^C. A.k.a. ‘it's a dry run, mother*****’.
<nckx>kmicu: Not Trisquel?? 😮
<g_bor[m]>how can I specify one of the outputs of a multi output package as input?
<kmicu>(I didn’t port my Gnubee to anything other than Debian for now.)
<nckx>g_bor[m]: ("foo" ,foo "out")
<g_bor[m]>nckx: thanks
<kmicu>I was too late to answer arkhan :/
<nckx>10 minutes is nothing.
<kmicu>We could at least say hi (╯︵╰,)
<Marlin[m]>hi guix
<nckx>kmicu: I'm never sure what's best. I'm already the ‘well I don't use GDM, but…’ person, it's not a great feeling to make noise without having much to say.
<nckx>I do try to say hi when I'm not busy.
<rubic88>Still no luck with i3lock. diff of my config.scm:
<xavierm02_>rubic88: Here's mine where i3 works:
<kmicu>rubic88: my guess is that your i3lock from specification->packages shadows i3lock from the service which has setuid set.
<xavierm02_>(Note: As is it makes sddm and gdm fight, so remove the gnome package if you use it)
<rubic88>xavierm02_: Thanks for the link.
<xavierm02_>rubic88: try removing uninstalling i3lock as a user maybe
<rubic88>xavierm02_: +1 That did the trick. Thanks!
<kmicu>How’s the status quiliro`?
<quiliro`>kmicu: i am with arkhan now
<quiliro`>he rebooted
<quiliro`>he did not login again
<quiliro`>i will reboot now
<quiliro`>arkhan is installing circe to connect again
<nckx>sneek: later tell arkhan: Hi! ☺
<sneek>Got it.
<rekado>civodul: we never actually fixed the general problem, namely that stale state files can cause GDM / gnome-shell to fail.
<g_bor[m]>hello guix!
<g_bor[m]>I have a package that needs gcc 6+, and I have read the issue about it, and it is still open. I'd like to know if pushing a package with a workaround is acceptable until it is resolved, or should I wait?
<g_bor[m]>(standard headers are not found)
<dGtlc3>Is gdm default lm for all de variants?
<mbakke>g_bor: Just remember to remove the workaround once it is no longer needed :-)
<rekado>dGtlc3: there are no default DE variants. The GDM service is included in the list of default desktop services, but you can easily edit that list of services.
<g_bor[m]>mbakke: ok, I instrumented it with some todo comments, and also made a note.
<dGtlc3>Was just wondering if it comes with all DEs. Thanks :)
<quiliro> again
<quiliro>kmicu: worked!
<quiliro>just that my keyboard is not in spanish
<kmicu>Epic saga is finally over. 🎉
<quiliro>kmicu: yes! how did you solve the video problem_
<quiliro>i still have to solve a couple of things> keyboard layout on i3, fonts for icecat
<zacts>does guix still rely on any Nix code, or is it all guile scheme now?
<zacts>(just curious)
<zacts>I'm just trying to understand if guix is implemented as a toolkit on-top of Nix, or if it is a reimplementation inspired by Nix
<zacts>I might switch over to Guix from NixOS eventually.
<kmicu>quiliro: the rest is much easier to fix and more related to Guix. So no worries.
<quiliro>how can i solve those two problems?
<kmicu>zacts: ‘a reimplementation inspired by Nix’ is an acceptable desrciption.
<quiliro>kmicu: thank you very much for your help with the video...have spent a week!
<quiliro>kmicu: so it was the kernel?
<kmicu>quiliro: to fix fonts in IceCat install e.g. font-gnu-unifont, font-dejavu, font-lato and execute ‘fc-cache -r’. After that restart IceCat and fonts should work.
<quiliro>where is that on the manual?
<kmicu>quiliro: yep, a recent kernel update changed things related to Openchrome support. For now you can stay on 4.19 and comment out ‘(kernel …)’ line in the config from time to time to check whether the issue is fixed in the latest kernel.
<quiliro>kmicu: how can i solve the keyboard issue? it is correctly defined on the config file...i checked twice
<kmicu>quiliro: I didn’t investigate why IceCat doesn’t work out-of-the-box but I installed some fonts, refreshed cache with fc-cache -r and that fixed the issue for me.
<kmicu>quiliro: I think the problem with ES layout is with my switch to slim. Let’s go back to GDM.
<quiliro>ok..hgow do i do that?
<quiliro>let me please.,...
<quiliro>you tell me if i am correct
<zacts>can I build local packages with plain ./configure;make on guix?, or how does this work?
<quiliro>kmicu: remove "remove"
<kmicu>quiliro: check if ES layout works with I removed SLIM.
<kmicu>(Basically going back to your old config. Though with (kernel …) line stays to fix Openchrome issue.)
<zacts>Like, if I wanted to build a user local perl or ruby, how might I do this on GuixSD?
<zacts>say a development version of perl or ruby
<quiliro>zacts: make a new guix package for that version
<zacts>quiliro: ok, cool
<quiliro>it is not too dificult
<rekado>zacts: yes, you can do this, but it would be easier to build Guix packages instead.
<zacts>that's cool, I see.
<quiliro>kmicu: i think i pasted the config file with an there a way that i download it to a my machine?