IRC channel logs


back to list of logs

<rekado_>I mentioned info.js, not node.js.
<rekado_>info.js is an implementation of standard Info reader features (keybindings to access the index, to search, to navigate between nodes) for the HTML output Texinfo sources.
<daviid>ah, ok
<daviid>i wonder how much work it would be to have scribble in guile
<civodul>rekado_: is it the one mthl worked on as part of GSoC?
<rekado_>the styling isn’t pretty, but with some tweaks it really makes navigating the manual much more enjoyable.
<rekado_>we’re using texinfo for internal sysadmin documentation at work, and with info.js my colleague (who really doesn’t want to use Emacs for some reason) can finally see the benefit of a good index.
<civodul>ah that's nice
<civodul>it saddens me when people just don't know about these manuals, or browse them sub-optimally
<rekado_>I remember that back then on Arch they aggressively deleted all info manuals, because they didn’t like the.m
<rekado_>“man pages rule”
<jonsger[m]>ACTION had started today reading the guile manual and understands now some stuff in guix even better :)
<buenouanq>for some reason my openssh service isn't starting automatically
<buenouanq>any idea?
***Apteryx_1 is now known as apteryx
<apteryx>rekado: that's shocking
<apteryx>My life's been so much better since Guix made me learn about how to use info ;). And I curse systems which don't ship with the info manuals.
<buenouanq>I get things to work by pulling from examples, but I still don't really understand the differences between use-modules, use-service-modules, and use-package-modules and what is supposed to go in them and how I find out.
<buenouanq>like, are (use-modules (gnu services networking)) and (use-service-modules networking) equivalent?
<iyzsong>Yes, you're right. See the end of the file 'guix/gnu.scm'.
<buenouanq>where is that file?
<iyzsong>if you have run 'guix pull', it can be found at ~/.config/guix/latest/gnu.scm. You can also git clone the guix repository.
<iyzsong>it can also be found at /run/current-system/profile/share/guile/site/2.2/gnu.scm in a GuixSD system.
<buenouanq>neat, thank you
<apteryx>buenouanq: the clearest is to only use the Guile builtin use-modules. The others are Guix specific syntactic sugar which might confuse more than help.
<apteryx>(in particular, they confuse Geiser which makes it harder to debug)
<buenouanq>well, having just discovered guix package --show=thing | grep location, at least one of my major questions has been answered.
<buenouanq>what's the equivalent for services or things in gnu/system/ though?
<buenouanq>like the nss-cert example says it needs (gnu system nss)
<buenouanq>but how would I find that out if it didn't tell me?
<iyzsong>system services can be searched using 'guix system search', and the "Services" of info manual document them.
<zero21>GLib-GIO-ERROR **: Settings schema 'org.gtk.Settings.FileChooser' is not installed
<zero21>how to fix?
<buenouanq>zero21: upgrade maybe
<buenouanq>I had that problem with audacity for a minute but I think it went away.
<zero21>i guess it wont went away, i thought its just drracket problem, now i get same with libreoffice they require this thing but apparently it is not present or cannot be found
<buenouanq>mine has been fixed
<buenouanq>pull and upgrade
<buenouanq>possibly even system reconfigure
<zero21>as root or user?
<buenouanq>root first prolly
<zero21>ok thanks
<buenouanq>hope it works
<iyzsong>I'm afraid it won't works... both packages need to be fixed. For now you manuall can add the path of '/gnu/store/gtk+-xxx/share' to XDG_DATA_DIRS.
<buenouanq>shepherd: Service user-homes could not be started.
<buenouanq>I think I've got this error every time since I first tried GuixSD - What is it, what does it mean?
***danialbehzadi1 is now known as danialbehzadi
<buenouanq>I cannot figure out why the ssh-daemon won't start on boot. I can start it manually after, but it's not being pulled up as the service I want/need.
<buenouanq>different issue, but the clock also refuses to be the correct time for some reason
***danialbehzadi1 is now known as danialbehzadi
<g_bor>Hello guix!
<g_bor>I just had a problem tonight, guile-2.2.3 not building, it just stopped after starting test.
<g_bor>I'm rebuilding to see if this persists. Anyone else experienced this? This is on core-updates.
<buenouanq> some of the modules I probably don't need
<buenouanq>and this builds without error,
<buenouanq>but ssh-daemon will not start on boot and I've no idea why.
<buenouanq>There are other weird things going on with gnome too, so I'd love another set of eyes to check if I'm missing anything.
<buenouanq>and if indeed isn't on my end, how do I determine if it is a GuixSD problem or a Gnome problem?
<wigust>buenouanq: What are those 'weird things'?
<buenouanq>the clock refuses to change
<buenouanq>hang on the Logging in... screen for a really long time sometimes
<wigust>buenouanq: I guess everything in desktop environment things which modifies system state (require root user privileges or dbus) will not work or will be temporal.
<wigust>buenouanq: You have a timezone in your system declaration. This is a way you want to change clock settings.
<wigust>buenouanq: What does 'sudo herd status ssh-daemon' after reboot say?
<wigust>buenouanq: I have a working sshd, but without X11 forwarding setting.
<buenouanq>that it's stopped
<buenouanq>I can start it and it goes just fine, but that's not the point in having it be a service...
<buenouanq>it even says it's to be respawned or whatever
<buenouanq>so then why is it stopped?...
<buenouanq>I've tried without the X11 forwarding, same deal.
<buenouanq>Gnome seems to just ignore some things sometimes - I have the timezone set in the config, it doesn't seem to care.
<buenouanq>basically the same config on an older install and everything is working.
<buenouanq>two latest installs don't pull up the sshd
<buenouanq>so it makes me think this is not on me
<buenouanq>but if I'm the only one complaining about it...
<wigust>buenouanq: Do you mean reconfigure or install?
<catonano>I issued a system reconfigure and now Inkscape is being compiled. I don't see how Inkscape can be among the dependencies of anyting
<buenouanq>wigust: three different physical machines
<wigust>buenouanq: So did you use 'reconfigure' or 'init'? I think most of people use 'reconfigure' and cannot spot after install failure. Still we test every change in service and use 'guix vm' with sshd.
<buenouanq>machine that works hasn't been reconfigured in a couple months
<buenouanq>the two I installed from scratch this week have this daemon problem
<wigust>buenouanq: What does /var/log/shepherd.log say?
<buenouanq>there's like an ordering problem
<buenouanq>things try to start before the things the require are happening
<buenouanq>some recover, but sshd does not
<buenouanq>this one actually doesn't illustrate what I mean
<wigust>buenouanq: Don't you have /var/run/ after reboot without starting it manually?
<buenouanq>dmesg has a few repeats of `enp3s0: link down' at the end
<buenouanq>the very last it says link up and the link becomes ready
<wigust>buenouanq: Did you you 'guix pull' before installation?
<buenouanq>I think on the laptop I did, but not on the second desktop.
<buenouanq> here's shepherd.log from the laptop
<wigust>buenouanq: I believe that 'guix init' will make you a different system (newer) if you made a 'guix pull'
<buenouanq>so is that something I should or shouldn't do?
<buenouanq>they told me yesterday to not pull prior to install and that it shouldn't matter because I can just pull and reconfigure after
<wigust>buenouanq: 'dbus-system could not be started' in your log
<wigust>buenouanq: dbus-system -> networking -> ssh-daemon
<buenouanq>right, that was the tail chase I was talking about
<wigust>buenouanq: I think they told you because you will build a released version commit, so 0.14 or what ever
<buenouanq>network and things are still able to recover, but sshd isn't
<buenouanq>question is, regardless of what happened during install, why doesn't a pull and reconfigure straighten it all out?
<wigust>buenouanq: I have no any dbus-system error in my log from 2017 october.
<buenouanq>this was installed fresh yesterday
<buenouanq>ntpd is constantly having to respawn now that I'm watching the log
<wigust>buenouanq: Could you try to herd 'status' and then 'start' dbus-system?
<buenouanq>status of sshd is
<buenouanq>dbus did get up an running in there
<catonano>now it's bilding Nautilus
<buenouanq> dbus-system status
<catonano>hi ng0 !
<ng0>would anyone like to take over the neomutt update? I can fill you in on details about the build system change
<ng0>I just don't have the time to do it
<ng0>ideally we want to unbundle autosetup and replace it with our own version
<ng0>as autosetup is meant to be included
<ng0>I have a wip branch for this somewhere
<wigust>buenouanq: Sorry, I cannot help more. You could ask on mailing list by sending an email to I believe the issue is in dbus-system failing to start.
<wigust>buenouanq: It should probably work for a release version you installed without 'guix pull'.
<buenouanq>So I should reinstall from the image, and not pull/upgrade/recorfigure?...
<wigust>buenouanq: As folks recommended you, yes. You could 'guix init' and if needed 'guix reconfigure', but don't pull for the debugging purpose.
***pksadiq_ is now known as pksadiq
<wigust>buenouanq: Or if ignore it and do 'guix pull', then you should show your current guix commit by 'guix --version'
<wigust>buenouanq: So people could reproduce it.
<buenouanq>wigust: I guess I'll be reinstalling it again this week.
<buenouanq>just from the website download and without pulling
<catonano>the Languagekit project targets llvm. Is it good for Guix ? And how does that relate to the GNU project ?
<wigust>buenouanq: Also be sure that image downloaded correctly by verifying signature as says. And be sure usb flashdrive is good.
<wigust>correctly: This is bad for user's freedom for sure (non-copyleft thing), but we have in Guix packages with those licenses.
<adfeno>catonano wigust: Agree, LLVM is one of the worst indeed.
<adfeno>The question is: what should we do? (a) Provide recipes to build with GCC; (b) Use LLVM.
<adfeno>I vote for (a), but it would be somehow more work in the begining. After the recipes are ready, however, we will have the option to send them upstream. This is an option of course, not a requirement.
<adfeno>... but I wish that LanguageKit would soon figure out the right reasons for using GCC build recipes/instructions instead.
<adfeno>... before something worse happens.
<catonano>adfeno: something worse ?
<critter13>Hi everyone! I was installing guixsd on my new libre x200 laptop, and I cannot figure out how to get wifi to work on the installed system. Is there a service I have to enable?
<critter13>I can get wifi to work on the live installer disk btw
<critter13>I am currently only using (services %desktop-services)
<snape>critter13: you could install wicd or cmst
<snape>cmst is the GUI client for connman (service connman-service-type (connman-configuration))
<critter13>snape: i think they are already installed... I guess my main question is: where do I put my wifi SSN and password?
<snape>critter13: what I do is: I run the GUI client, and type my password there
<critter13>snape: and how do I get the wifi enabled at startup?
<snape>do you use wicd-service or connman-service (or something else?)
<critter13>snape: networkmanager GUI?
<snape>yeah for example
<snape>I think it's GUI is nm-applet
<critter13>snape: idk, im coming from arch linux with netctl
<snape>I don't use networkmanager
<critter13>what do you recomend?
<snape>either wicd-service or connman-service. None of them are perfect though
<snape>wicd-service doesn't allow for several connections are the same time if I recall well
<critter13>snape: alright thank you. I will rtfm now
<snape>I just realized that &desktop-services includes Wicd
<wigust>critter13: I believe you could setup wifi with nmtui or nmcli with networkmanager provided by default %desktop-services
<critter13>wigust: ok
<snape>so, critter13, you could just install the wicd package, and run wicd-gtk
<snape>this will give you a GUI where you can browse the wifi and type passwords
<wigust>snape: We switched to networkmanager in %desktop-services some time ago, didn't we?
<snape>wigust: the docs doesn't say so
<snape>it should be updated then
<adfeno>catonano wigust: Re: LangugageKit and GCC vs. LLVM: Something worse in terms of "software freedom not being available for the end-user".
<adfeno>.... aaand I can't type today... ;)
<wigust>catonano: See about llvm from RMS
<snape>wigust: critter13: I was reading this: sorry
<adfeno>catonano wigust: I may not be an experienced programmer, but I can already see an issue here if LanguageKit and other projects keep up with this madness to support LLVM.
<critter13>one more would I have ntp run on startup as well
<critter13>it thinks im in 1970 every time I boot
<catonano>adfeno: we should probably move this discussion on the mailing list. I'd love having a deeper discussion about this
<wigust>critter13: it should be already started up, but you need a network connection to sync. 'sudo herd status ntpd' to check
<wigust>critter13: to check did it start
<snape>I think there are issues specific to x20 hwclock
<catonano>wigust: Stallman's assessment settles the point of the relationship between the GNU project and LanguageKit
<catonano>it´s a pity because otherwise GNUStep is an interesting project
<adfeno>catonano wigust: Agreed to move to mailing list. ;)
<adfeno>... who will do the move?
<catonano>adfeno: I will ;-)
<adfeno>Oh OK thanks catonano ;)
<snape>critter13: I remember I did something like this:
<snape>maybe it'll help, dunno
<critter13>snape: thanks I will look into it furthur
<rekado>critter13: I have the same problem with my x200s and latest Linux.
<rekado>previous versions of Linux didn’t trigger this clock problem.
<snape>rekado: 4.12.13 only?
<snape>4.14.13 I meant
<catonano>adfeno: I wrote on the dev ml
<adfeno>catonano: OK ;)
<adfeno>Thanks! :D
<g_bor>Hello guix!
<g_bor>I just noticed that nspr 4.17 on core-updates is not reproducible.
<g_bor>Anyone has any information about that?
<g_bor>-ELF>`@@UH��H�H�E�H�H�E�H�]�$Header: NSPR 4.17 (debug) 2018-01-13 14:27:53 $@(#)NSPR 4.17 (debug) 2018-01-13 14:27:532018-01-13 14:27:53libnspr4.soPortable runtimeN/ACopyright (c) 1998 Netscape Communic2@+ons Corporation. All Rights ReservedLicense information:�/�bGCC: (GNU) 5.5.0zR�#A�C
<g_bor>+ELF>`@@UH��H�H�E�H�H�E�H�]�$Header: NSPR 4.17 (debug) 2018-01-13 14:27:45 $@(#)NSPR 4.17 (debug) 2018-01-13 14:27:452018-01-13 14:27:45libnspr4.soPortable runtimeN/ACopyright (c) 1998 Netscape Communic2@+ons Corporation. All Rights ReservedLicense information:�;/�bGCC: (GNU) 5.5.0zR�#A�C
<g_bor>It seems like these two lines represent the problem.
<g_bor>The difference is the additional ; in the second row.
<ng0>setting up setuid programs in '/run/setuid-programs'...
<ng0>guix system: error: copy-file: No such file or directory: "/run/setuid-programs/xscreensaver"
<ng0>has anyone tried xscreensaver before?
<ng0>I rememeber nothing, and thought that 'xscreensaver' needs to be suid
<ng0>because of:
<ng0>xscreensaver: 17:07:09: warning: /etc/pam.d/xscreensaver does not exist.
<ng0>xscreensaver: 17:07:09: password authentication via PAM is unlikely to work.
<ng0>wee. same problem for slock now
<ng0>could be that this is due to offloading.
<ng0>ACTION tests
<atw>ntp-service should be able to adjust my system clock, right?
<atw>The behavior I am seeing is shepherd respawning ntp every 3 minutes or so, and the time not being adjusted
<catonano>sometimes I whish Guix developent was not so fast. Pulling, reconfiguring and building VMs I have been compiling and building stuff most of today ;-)
<lfam>If you don't keep pulling, then you don't have to rebuild ;)
<ng0>oops. solved the "issue"
<catonano>lfam: right. I alway overlook pulling. I'm learning slowly
***Piece_Maker is now known as Acou_Bass
<castilma>do i have to run ./bootstrap&&./configure --localstat... &&make after every git checkout/pull?
<atw>from my /var/log/messages:
<wigust>castilma: no, only if you never did it as after git worktree or git clone
<castilma>so, after a checkout, i can directly use guix?not even make necessary?
<castilma>i somewhere got the impression that at least make or so is necessary. thanks for clearing that up.
<wigust>castilma: only 'make' after checkout/pull, sometimes 'make clean' if abi breaks
<castilma>wigust: thanks. another minor (?) issue i have. i installed guixsd on my harddrive to dualboot into it. currently i run nixos and booted guixsd with qemu. i try to run the make check on 138c088, but some test makes guixsd freeze. it doesn't seem to respond to anything anymore. how can i debug that?
<wigust>castilma: sorry, missed, no 'make clean' but 'make clean-go'
***Piece_Maker is now known as Acou_Bass
<wigust>castilma: i believe it runs test in qemu
<wigust>castilma: so you are running qemu inside qemu
<castilma>and that doesn't work?
<wigust>castilma: i think it should work, but i didn't do it
<wigust>castilma: does it work if you boot with your dualboot and test from there?
<castilma>wigust: didn't try that, jet
<atw>What are the arguments for (guix store)'s build? I would like to run it on an expression in the store-monad repl, something like (build (gexp->script "foo" my-gexp))
<lfam>efraim: Should we remove armhf from Go's supported-systems? Has it ever worked on armhf?
<ng0>I just offloaded the build for texlive after a year without offloading. Now I do understand why we don't build substitutes for it :D
<ng0>waiting for it to finish even in one local network is like waiting for the drum beat in drum less drone music.
<efraim>lfam: its supposed to build, but I don't have access to an armhf machine to work on it
<lfam>efraim: Do you think I should just remove armhf from supported-systems for now? To save Hydra some effort and unclutter the failed builds list some?
<efraim>For that we could ditch Java on non x86_64 platforms
<efraim>I'd rather leave it there, or remove it if we ever get a gccgo package working
<lfam>We could always put it back later if somebody wanted to work on it. Although I wonder how much interest there will be in armhf in the future. armv8 is already cheaper and much faster
<lfam>I wish there was a Hydra-specific way to avoid building some packages. Using Guix stuff like supported-systems to have an effect on Hydra is really not the right way to do it
<ng0>hahaha... still not fully received the build
<ng0>ACTION makes note that this was a bad idea
<amz3>does anyone want to work on guix publish using gnunet?
<amz3>now that they are bindings, it shuld be more easy
<ng0>I mean, yes, here. I can and want to help
<atw>ng0: I'm not very familiar with offloading...would it have been better if you had done this locally? just curious
<ng0>yes, because texlive is quiet large
<ng0>we have disabled substitutes for this on the build farm
<atw>what would happen if you disconnected from the network while waiting to receive an offloaded build?
<ng0>iirc it would just time out after a while
<bkeys>Can Guix build packages from source?
<bkeys>What is the option?
<lfam>bkeys: You can use --no-substitutes, which is a common build option:
<pkill9>i dunno
<bkeys>lfam: So that will build the package from source?
<pkill9>i'd like to use guix, i like the look of guile
<lfam>Actually, Guix is a build-from-source distro. The binary substitution is just an optimization made possible by the functional package building model
<bkeys>I am just trying it out in a VM at this point; I have used Fedora for a long time
<pkill9>i use nix as a package manager, but i would much prefer guix as the package definitions seem much nicer to read, and i don't think i would need to learn much guile before being able to modify them and make new ones
<pkill9>i only use nix because I use proprietary nvidia drivers, if i build new pc i'll get AMD card for open source drivers
<bkeys>I am personally in the process of limiting the nonfree software to just nonfree firmwares
<bkeys>Soon I will have nothing but ARM based computers with a workstation with an AMD GPU
<bkeys>Fedora supports ARM and all that; but right now my work VPN only supports Ubuntu, my boss said it's likely cause the VPN hates anything with systemd
<bkeys>So I am getting a chromebook for a work computer and since Guix doesn't use systemd I want to see if I can get it to connect to the work VPN once the chromebook comes in
<bkeys>Cuase i needed it to not have systemd, and support ARM at the same time
<pkill9>i use slackware atm
<pkill9>that doesn't have systemd
<pkill9>which is nice
<lfam>Currently GuixSD doesn't run on ARM. We would like to make it possible, but it hasn't happened yet
<pkill9>i don't like systemd
<lfam>Guix the package manager does work on armhf and aarch64 on top of another distro
<pkill9>there's an ARM version of Slackware
<bkeys>lfam: You can download ARM binaries here though
<lfam>lfam: Yes, for the package manager
<pkill9>Guix != GuixSD
<pkill9>Guix = package manager, GuixSD = Operating System that uses Guix
<pkill9>although 'uses' is an udnerstatement, the whole system is managed by guix
<pkill9>including configuration i believe
<castilma>can someone test sth for me in guixsd? what does `whereis suspend` give you, and does bash tabcomplete `susp` for you?
<atw>castilma: whereis suspend just prints "suspend:" for me. Bash tabcompletes susp to "suspend "
<castilma>atw: thanks. and if you execute it, does your bash sleep indefinitely and guixsd stay awake? (warning: you may not be able to kill bash except with SIGKILL, if you test this)
<atw>castilma: by "guixsd stay awake", I assume you mean "the computer does not suspend-to-ram"?
<atw>ok, I will try it. I may be disconnected briefly.
<atw>Just started it as my normal user and no suspend yet.
<atw>OK, ^C doesn't stop it and I can't kill the xterm either
<atw>a SIGTERM didn't kill it, a SIGKILL did.
<castilma>so, what is bash trying to do?
<castilma>you are not using xfce, right?
<atw>I am using spectrwm. If you want to suspend, I think "loginctl suspend" might do it for you. Maybe you could try to spy on the process with strace or gdb or something...
<alezost>castilma: atw: "suspend" is the shell built-in command as you can check with "type suspend"
<apteryx>Q: Do we still want the build phases to return #t upon success? Or have we moved to the exception based system discussed on the mailing list?
<ng0>it wasn't communicated very clearly outside of the thread.
<atw>alezost: thanks for clearing that up suspend suspends bash itself until receiving a sigcont. Should have looked that up earlier :P
<apteryx>ng0: OK. I'll keep to my old habits for now.
<ng0>I meant to express: I would like a reply as well
<apteryx>Is there a way to sort bugs in a decreasing importance manner in Debbugs? I found the 's' toggle but is sorted in increasing importance.
<apteryx>ooh. I could... click on it.
<apteryx>(on the State column, that is).
<Apteryx>any suggestion for parametrizing a build phase?
<Apteryx>I'm thinking about a new emacs 'make check' phase, except the test target should be parameterizable as it's not always 'check'
<Apteryx>Maybe I can simply add an argument to the emacs-build-system.
<Apteryx>oh, yes that will do, it's already implemented for the gnu-build-system as the `test-target' argument.