<lyr3>hey, after installing some packages with (guix package -i) I get two errors: <lyr3>unsupported manifest format and some backstrace output <lyr3>I wonder what I am doing wrong <mange>I think "unsupported manifest format" is when the profile was built by a more recent, and incompatible, version of guix than what you're currently using. At least, that's where I've seen that before. <mange>It's hard to help with the backtrace without actually seeing it. Can you make a paste of it? <lyr3>Too bad, I will reinstall and redo all of it again. <lyr3>About the unsupported manifest, that happened always after a guix pull (user) <mange>So you ran guix pull as a regular user, then guix package -i failed as that user? <lyr3>that last time, I did a guix pull as root, then a guix pull as user <lyr3>guix package -i did recommend to guix pull and guix package -u <lyr3>after installing the first package <lyr3>How do I install package for all users? ***dmc is now known as Amazon
<mange>Add it to your system's config.scm file, in the packages field of your operating-system. <lyr3>oh, guix reconfigure. So what is the usability of the local guix package -i under GuixSD <mange>On my GuixSD machines I have very few packages installed system-wide (I have nss-certs, and whatever's in %base-pcakages), and I install practically everything as user packages. <lyr3>And how do you update packages installed through guix reconfigure xx.scm? <mange>sudo guix pull && guix reconfigure xx.scm <mange>You have to update the whole system to update the system-wide packages, which is a good reason to favour user packages. <mange>Well, maybe not a good reason, but a reason. Usually it's quicker for me to update my user packages than to reconfigure my system. <lyr3>There is any way to install xfce or wm as user and login? I could not achieve that <lyr3>Slim wont reckon wm/de installed by users <mange>I think Slim will still run your .xsessionrc, so you can source ~/.guix-profile/etc/profile (to get $PATH, etc.), then run your window manager of choice. <mange>I have (xfce-desktop-service) in my system services, though, so I use the system xfce. <lyr3>thank you. Time to reinstall it! haha ***Amazon is now known as dmc
<Apteryx_>where is the system's shepherd config stored on GuixSD? <Apteryx_>I guess it's only kept in the form of the derivation of the config.scm file that describes my system? <mange>It's stored in the store. I think it's /gnu/store/...-shepherd.conf, but finding out which one corresponds to your current system is trickier. I can't remember how to do that. <mange>Maybe it's somewhere in /run/current-system/? <baconicsynergy>will report anything I encounter, and possibly any suggestions to improve the instructions <bnw>I deleted /etc/ssl/certs/ca-certificates.crt since I thought I was from some unknown source. Then I found out it probably was from guix (running in Debian). Where can I get it back? <marusich>bnw, just run "guix system reconfigure" on your latest config file. <marusich>That file is auto-generated by a profile hook, so when you do that, it will recreate it for the system profile. <baconicsynergy>sneek: tell roptat /system/etc/ and /etc/ are both present on lineage 15.1. which should I choose? <mange>I don't think that file was from Guix. If you're running debian, I'm fairly sure debian puts that file there. <marusich>bnw, my advice applies only for GuixSD; if you're on a foreign distro, I don't know what to do. <mange>I'm currently on a debian machine, and I have an /etc/ssl/certs/ca-certificates.crt file that I'm at least 90% sure Guix didn't put there. <marusich>There is a way to regenerate the file on Debian <marusich>I don't remember exactly, but if you do an Internet search I'm sure you'll find it. <bnw>I saw almost all files under /etc/ssl/certs/ were soft link and thought normal files there were from unknown source <vagrantcish>bnw: apt --purge --reinstall install ca-certificates, probably <bnw>apt --reinstall install ca-certificates did the trick <jeko>I try to play withh guixsd but i have troubles to setup my vm network... <snape>Apteryx_: if you 'ps auxww | grep shepherd', you'll find its configuration file <g_bor[m]>jeko: so, you are using the installation image? <jeko>the GuixSD 0.14.0 QEMU Image <g_bor[m]>jeko: do you receive any error messages? <jeko>g_bor[m]: when I ran this command 'ps auxww | grep shepherd' did get an error message <snape>jeko: I was not talking to you actually :-) with IRC, if jane says: "john: something", it means jane is talking to john. <g_bor[m]>jeko: ok, just to clarify what is already done, did you run 'ip a', what was the result, and what is the name of your primary network interface? <snape>jeko: Also, saying someone's nickname allow their IRC client (well, most of them) to beep, so they get notified that someone is talking to them. <jeko>snape: right, sorry haha <jeko>g_bor[m]: "ip a" or "ifconfig -a" give me lo and ens3 <jeko>g_bor[m]: then I did "ifconfig ens3 up" <jeko>g_bor[m]: then "dhclient -v ens3" <jeko>g_bor[m]: the result is inet addr : 10.0.2.15 for ens3 <g_bor[m]>jeko: And after that, pinging a public ip does not work? <g_bor[m]>jeko: is that ip on the expected network? <jeko>g_bor[m]: exactly the ping result in only transmitted packets and 0 received <jeko>g_bor[m]: how can I know which network is used? <g_bor[m]>jeko: ok, it seems that the problem is not on the guixsd side. Do you have any other vm with a working network? How do you start your vm? <efraim>Ping is known to not work in the vm <jeko>g_bor[m]: I have nothing to compare actually :/ the command I use to start is "qemu-system-x86_64 -net user -net nic,model=virtio -enable-kvm -m 2048 guixsd-vm-image-0.14.0.x86_64-linux" <jeko>efraim: me neither hahaha <g_bor[m]>efraim: do we mention that in the documentation? Should we suggest another way to test network in a vm? <jeko>Alright then I tried guix buid hello and the network is clearly the issue... "ERROR: In procedure getaddrinfo: Name or service not known" <jeko>when trying to download binutils <g_bor[m]>This might be a name resolution problem. I am not sure what is available on the vm image. What is the output of 'ip route show'? <g_bor[m]>jeko: you might try to add a nameserver there. <jeko>g_bor[m]: ip route list returns nothing <jeko>g_bor[m]: in resolv.conf I found "nameserver 10.0.2.3" <g_bor[m]>jeko: it seems that routing information is somehow missing. What is the output of ip addr show ens3? <g_bor[m]>jeko: also, can you ping 10.0.2.2? (this ping should work) <numerobis>Hello guix! I've got another question regarding build reproducibility. Suppose that I wrote a small program X, that depends on a series of libraries, and that I want the results generated using program X to be reproducible. If I write a guix package for it, the result of the build will still depend on the guix distribution, right? (i.e., doing 'guix pull' might change the resulting executable, if the <numerobis>derivations of some of the dependencies change.) Is there a way to specify a git commit of the guix distribution in the package definition, or something like that? Or does the goal of reproducibility only apply to packages in the guix tree? Thank you! <jeko>g_bor[m]: Alright I found the problem ! <jeko>g_bor[m]: the reason was, I'm dumb... <jeko>g_bor[m]: between tries the network interface changes and i lost the seam <jeko>g_bor[m]: now I can download part of hello but there is a disk space issue. i think i will be able to manage it now <jeko>g_bor[m]: thank you very much for helping me <jeko>g_bor[m]: and... ping doesn't work in vm haha ! <g_bor[m]>jeko: you are welcome! I have also run into disk space issues earlier, the image should be resized. I did not realized the limitation of ping, it is not the limitation of vm, but the qemu user mode network. I'm not affedted, as my vms run on bridged network. <g_bor[m]>jeko: I have also used a customized installation medium to provision my vms recently. <g_bor[m]>jeko: I'm glad, that you finally could get it working. <jeko>g_bor[m]: have a nice day, I leave the channel! thank you again <dustyweb>I got a usb-c hub but it doesn't work and it looks like maybe a bluetooth driver is being kicked on and failing to load, though that makes no sense <dustyweb>I'm gonna try with not-linux-libre and on morgan's debian laptop to see if it works <efraim>we don't build the html output of the manual with the standard 'make all', do we? <Rukako>I don't know anything about it myself though <g_bor[m]>PotentialUser-27: I have not tried it myself yet, but is should without problem. There are some issues regarding documentation, though. <lfam>efraim: To build the HTML manual, I do `make doc/guix.html` <kmicu>PotentialUser-27: do you contribute to Pijul by any chance? :P <Rukako>should I prefer classes, records, or what in guile? <Rukako>I only used the racket structs and message passing (in the way shown in SICP) in the past <Rukako>I would just use a hash table or a list and define different "getters" that use caddr and stuff on it but I would like to know the "proper" way to do it in guile <Rukako>(especically when considering that guile supports SRFI-9, its own, as well as R6RS records) <catonano>Rukako: you may want to ask this question on #guile