<joshuaBPMan>hey, on guix system gpg --full-generate-key fails with "no pinentry" <reepca>joshuaBPMan: have you set up your ~/.gnupg/gpg-agent.conf? <joshuaBPMan>pinentry-program /home/joshua/.guix-profile/bin/pinentry <reepca>IIRC it might be necessary to restart the agent or something like that for it to pick up the changes <nckx>joshuaBPMan: Just (p)kill it. <joshuaBPMan>hmmm. I'm having a hard time getting ssh-add to import my gpg key. <jgibbons[m]>The farthest I got in building guile-emacs is when it segfaults while compiling calendar/diary-loaddefs.el <jgibbons[m]>I have 32GB RAM and 17GB swap. I don't think memory is a problem. <jgibbons[m]>Could it be something wrong with guile-for-guile-emacs? <jgibbons[m]>Could I get an advantage if I change limits for the guixbuild group? <nckx>Segfaults aren't out-of-memory errors. <jgibbons[m]>guile-for-guile-emacs would be the one compiling the elisp. <nckx>Maybe, or with guile-emacs itself, or a combination of the two. <nckx>guile-emacs is dead, right? <nckx>last change Tue, 12 May 2015 <nckx>Not that we shouldn't support it, of course, but it seems hard to keep an 4.5-year old emacs from bitrotting⌠<jgibbons[m]>Time to try to make it work again? I'll see if I can find time to do nothing but hack at it for a week. But that probably won't happen until January. <nckx>Sorry, I shouldn't have used âsupportâ like that. I mean âaccept patches and wait a reasonable while to remove it after it breaks.â I think that reasonable while has expired unless someone fixes it soon, and that someone might have to be you đ <nckx>And it mentions segfaults. <nckx>Ancient abandonware is fine as long as it works and is maintained by someone who uses it, but this isn't the first time that guile-emacs has flirted with being⌠quite the opposite. Maybe it would be more at home in a channel instead of Guix proper. <nckx>But don't mind me, I'm an exceptionally opinionated prick tonight. <jgibbons[m]>I'll see what I can do. Don't remove guile-emacs or guile-for-guile-emacs for at least another year. <nckx>Hehe. Don't worry, I won't. <nckx>jgibbons[m]: Maybe file a bug (if only for your future self) with your current progress/error? I don't know whether janneke's still interested in it. <nckx>jgibbons[m]: If you don't mind me asking: why are you? ***modula is now known as defaultxr
<jgibbons[m]>This wouldn't be the first time I dug through garbage to make it work. I temporarily fixed the libre MMORPG deliantra client and server last year. <bdju>emacsomancer: launching next browser with full path seems to have the same problem still <bdju>guile emacs seems cool, I am happy to hear that someone might try to save it <bdju>and I think it would fit well with other guix stuff <bdju>still doesn't work, flashes white for a second and then closes <efraim>I wonder if there's a way to refer to openssl so that it is openssl on master and openssl-1.0 on core-updates ***jonsger1 is now known as jonsger
<coco>Yesterday, I encountered odd, non-deterministic behavior of 'guix system reconfigure'. I proceeded by basically creating a minimal working example of this behavior by doing the following: <coco>(1) I made a fresh installation of guix with a minimalist config.scm (a stripped-down version of the 'bare-bone' version provided on the installation medium). (2) I ran 'guix pull' (as root, there was no other user yet). (3) I made some changes to config.scm. (4) I ran 'guix system reconfigure config.scm' a couple of times. And this is the result: https://pastebin.com/pFMNn3m1 . <coco>That's a new profile every time I ran 'guix system reconfigure', and worse than that, it keeps downgrading the system (see the kernel version). I have no idea why it behaves this way. Why could that happen? I appreciate any hints, as I find that rather frustrating. <efraim>since it keeps on downgrading I assume that .config/guix/current/bin wasn't added to the $PATH and you weren't getting the benefits of guix pull <coco>efraim: interesting, what's the difference of having in the path? what would be there and what would it otherwise use? sorry, i'm a guix newbie <coco>hmm, unfortunately, /root/.config/guix/current/bin is in the path, actually <quiliro>coco: I think the most plausible reason is that the changes you made to config.scm changed the required operating system kernel version. You can check that and see if the same config.scm installs the same kernel. If it is not that way, probably it is a bug. If you can confirm this bug, please file a bug by writing an complete descrition by email to bus-guix at gnu dot org. <coco>quiliro: i don't think that the config.scm is the cause, for two reasons: (i) i ran it multiple times with the _same_ config.scm (ii) i only configured the bare minimum: bootloader, file system, one user, a dhcp client and sshd <kmicu>coco: could you share your PATH? Is that a fresh install or do you have an old .bashrc (or other existing files) in your home? <quiliro>I do not know how to check guix version generations. Why don't you check the same becavior by adding a user with the wheel group and confirming (with sudo guix system reconfigure config.scm)? <coco>i'm at this point not 100% sure whether /root/.config/guix/current/bin was in the path when i ran it yesterday, but it is now; so i will just 'guix pull' and 'guix system reconfigure' a couple of times now <coco>kmicu: about the path, i just notice now that $PATH is different when i sit down and log in at the machine itself, compared to when i log in via ssh. when i sit down at the actual machine, it is: /run/setuid-programs:/root/.config/guix/current/bin:/root/.guix-profile/bin:/run/current-sytem/profie/bin:/run/current-system/profile/sbin. when i log in via ssh, it is (i log in as user coco and then su to <coco>root): /run/setuid-programs:/run/current-system/profile/bin:/run/current-system/profile/sbin <coco>just to clarify, i always ran 'guix pull' and 'guix system reconfigure' as root; i have the user 'coco' only to ssh into it and copy paste the output for the pastebin <coco>kmicu: it is a fresh install; no old .bashrc, only touched config.scm <coco>efraim: logging in via ssh and then su to root, i get /run/current-system/profile/bin/guix <coco>for 'which guix', that is <efraim>coco: that's the guix installed in the system profile. If you use it for 'guix reconfigure' it'll downgrade the system <coco>efraim: that sounds like i totally misunderstand something here. alright, how should i do 'guix system reconfigure' then? <coco>i'll try 'which guix' at the actual machine (not via ssh) once the running 'guix pull' is done <coco>and also to clarify, i always ran 'guix pull' and 'guix system reconfigure' at the actual machine, not via ssh <coco>efraim: would that also explain why it downgrades the system repeatedly? <efraim>the big thing I've changed in my .bashrc was I changed the line starting [[ to if [ -n "$SSH_CLIENT" -a -z "`type -P file`" ] <efraim>but I don't really understand how bash and the different files interact <coco>me neither, actually; that's why i want to restrict my observations to the case where i log in at the actual machine, instead of via ssh <coco>i also don't understand the differences between doing something with sudo, after su and after logging in as root <efraim>for reconfiguring we generally suggest running 'sudo -E guix refresh ...' <coco>interesting. 'sudo -E guix ...' is nowhere in the manual <coco>ok, 'which guix' (when not logged in via ssh) gives /root/.config/guix/current/bin/guix <coco>i guess that looks good? <coco>will run 'guix system reconfigure' two times now and see what happens <efraim>coco: yeah, that's the right one <coco>just ran 'guix system reconfigure config.scm' three times, and this is the result (see generations 6, 7 and 8 at the bottom): https://pastebin.com/FtwHY7Ne Looks good! :-) <coco>not really sure what the problem was before, but i think the most likely explanation is path issues / wrong guix binary <coco>efraim quiliro kmicu: thanks for your help! hope that solves it for me <coco>just for my understanding: at what point is /root/.config/guix/current/bin/guix added, and when should it get added to the path? <coco>is it added after the first 'guix pull'? <coco>i see. does that guix pull modify bash_login / bashrc / whatever, and i need to source it then? <rekado>coco: Guix does not modify any file outside of /gnu/store <rekado>âguix pullâ tells you to modify PATH by yourself. <coco>rekado: oh, that's where my bad is... i see <disxsdx>hi, flash question: does Guix use wailand? *mbakke has a fix for next and epiphany failing under wayland <Chiliparrot>btw: what's up with wayland (in general) nowadays? I don't have the feeling X11 will be deprecated in the near future, and things can get quite complicated when you switch <desmes>Hey guix. Any Next user here? Can you reproduce video content? I get a "your browser does not currently recognize any of the video formats aviable", should I install some other package? <rekado>desmes: does it use gstreamer? If so, you should try to install the gst-* packages. <rekado>(this is a common question and we canât really improve the defaults here, so I think it would be good to add this to the cookbook.) <desmes>rekado: I actually don't know. But I'll try isntalling them <rekado>desmes: you may also need to set a few environment variables to make gstreamer find these plugins. <pinoaffe>hi guix! is there a way to configure hosts (what one would do in /etc/hosts in a "normal" gnu/linux setup), either on a systemwide or on a per-user level? <grumbel_> (hosts-file (local-file "/home/ton/guixsd/dotfiles/etc/hosts")) <quiliro>how can i know if my system configuration has some package....so i will not install it in my user....an example would be gst-plugins-* <quiliro>of course it would be better that it would be install on all users...but it would be nice to know how to do it <efraim>Also I wouldn't worry about double installing packages. I use the system packages as bonus or required for some functionality but it doesn't affect my package choice in my manifest <pinoaffe>same, can't really think of a case where it'd be detrimental <quiliro>I would like to know if a package is installed on the system not on the user...how can i do that? <quiliro>found emacs-guix's M-x guix-system-from-file <quiliro>Wow...I never saw nckx exit the chatroom! <nckx>quiliro: As much as I wish that were true, I cheat & use a bouncer. <nckx>Or did you just imply I have no life?? <nckx>Because that would be true. <quiliro>why would you not be in #guix and be in #guix at the same time for? in other words, what is a bouncer good for, not being noticed until you want to? <nckx>It (amongst other secondary features) saves channel logs for me and displays a conversation history when I log in, as well as saving personal /msgs that people sent in my absence (happens more often than you'd think). <quiliro>that is probably because you do not seem present <nckx>Oh quiliro, I don't even âseem presentâ when you're face to face talking to me. <nckx>Maybe everyone gets /msgs from strangers all the time, it just surprised me when I first started using IRC⌠<quiliro>so you are one of those that read cell phone messages while on a meeting with family on the dinner table? *quiliro will get a signal blocker for the dinning room! <nckx>quiliro: Don't have a phone. <nckx>So, what's the skinny on core-updates? <mbakke>nckx: only mariadb on armhf remains, AFAIK <mbakke>not sure what do about it, feedback welcome :-) <nckx>Right. I was just reading that. It's a niche use⌠but it's a bad one to break. <mbakke>I've run multiple replicated MySQL clusters and did not even know it could encrypt binlogs until now :P <nckx>I'm a postgres boy. TBH IDK if it's just a test failure or a real problem? <mbakke>nckx: it does look like a real problem. Having read the test, the extra data comes from "@@skip_replication" statements that are ineffective, somehow. <nckx>mbakke: Did you mark it as Major or did they? <mbakke>nckx: I think that was upstream. *nckx reconfigures some Important Servers to core-updates already. <jgibbons[m]>gnu talkfilters are different from debian's filter package. I'm packaging Debian's filter package, but could use a bit of help identifying licenses. I asked on help mailing list. <jgibbons[m]>It's mostly about the artistic license. Is the license in /usr/share/common-licenses/Artistic on Debian the version FSF says is too vague to be free? Not running debian. *nckx ssh's into a Debian system. <nckx>jgibbons[m]: Hm, it doesn't match the text in either the Clarified Artistic or Artistic 2.0 licences in (guix licenses). <nckx>Copied straight from Debian, gotta go. <rekado>is it the original Artistic 1.0 then? That would be non-free. *rekado installed guile-studio on master, got an error when running it <rekado>does this look familiar? /gnu/store/87mnipnpgwfry5lss785zxx197wz6pg5-emacs-26.3/bin/emacs-26.3: /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/libm.so.6: version `GLIBC_2.29' not found (required by /gnu/store/aap7x1likrb4837sajxmz1wgffql15zj-librsvg-2.40.20/lib/librsvg-2.so.2) <rekado>this is because gnome-shell is started with LD_LIBRARY_PATH <nckx>which is a non-free licence. <nckx>jgibbons[m]: âŚright? I'm afraid Kenny will need to go. <jgibbons[m]>I can contact the copyright holders and see if they can license under clarified-artistic or artistic-2.0. Until then, I'll kill Kenny and remove any reference to it from Jibberish. <quiliro>how to use SICP after guix install sicp? <rekado>quiliro: âC-h iâ in Emacs to view the info index; select SICP there. <jgibbons[m]>"This software is copyright (c) 2014 by Chris Williams, Christian Garbs and Alan Eldridge. <jgibbons[m]>This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself." <jgibbons[m]>Christian Garbs and Alan Eldridge are the authors of Kenny from filters. <joshuaBPMan>hello, I'm trying to use my gpg keys for ssh authentication. I am really struggling to do this right. Anyone have any suggestions? <joshuaBPMan>rekado: I suppose. I was hoping to use gpg-agent, but I suppose ssh-agent will probably work too. <joshuaBPMan>actually it might not. I'm using sway. I don't think sway executes ~/.xsession <rekado>I managed to make my system tests pass only when rpc.mountd is running under strace; they fail when itâs running without strace. <doubleloop>i get this error when installing guix: /gnu/store/...-grub-efi-2.02/sbin/grub-install: error: failed to get canonical path of '/boot/efi' <doubleloop>this is via the guided graphical installer, selecting the option to use the full disk <rekado>doubleloop: are you sure that you want to do an EFI and not a legacy BIOS installation? (Sometimes the installer gets confused with what the system tells it about the boot mode.)