Guix System

Guix v1.0.0 has officially been released! Step up to based beautiful future of package management, the GNU operating system, and free software in general. There's even a graphic installer now for you incapable retards. And they're just calling the OS `Guix System' now instead of GuixSD. I'm here to help. https://www.gnu.org/software/guix/download/ https://www.gnu.org/software/guix/manual/en/html_node/System-Installation.html

Is it systemd-less?

>>16957 Of course. It uses GNU"s Shepherd and everything is painlessly configured using Guile.

I like to install an extremely small instance and then build it up as I need to from there. You can discover any issues you might have quickly and don't have to wait for a ton of stuff to build before booting into the new install. Here's about as minimal as you can get: (use-modules (gnu)) (operating-system (host-name "desktop") (timezone "America/Denver") (locale "en_US.utf8") (bootloader (bootloader-configuration (bootloader grub-efi-bootloader) (target "/boot/efi"))) (file-systems (cons* (file-system (device (uuid "E4E4-AC21" 'fat)) (mount-point "/boot/efi") (type "vfat")) (file-system (device (file-system-label "a")) (mount-point "/") (type "ext4")) %base-file-systems)) (swap-devices '("/dev/sda3")) (users (cons* (user-account (name "buen") (group "users") (supplementary-groups '("audio" "netdev" "video" "wheel")) (home-directory "/home/buen")) %base-user-accounts)) (packages %base-packages) (services %base-services))

Have some gross hack aliases I've found useful: gpFi () # install all font packages { guix package -i $(guix package -s ^font- | grep name: | cut -c 7-) } gpG () # list globally available packages { INSTALLED="$(PACKAGES="$(guix package -p /run/current-system/profile \ -I | grep -v ^font-)" echo "$PACKAGES" | while IFS= read -r PACKAGE; do NAME="$(echo "$PACKAGE" | cut -f 1)" VERSION="$(echo "$PACKAGE" | cut -f 2)" OUT=":$(echo "$PACKAGE" | cut -f 3)" if [ "$OUT" == ":out" ]; then OUT="" fi echo "$NAME""$OUT" "$VERSION" done)" echo "" echo "$INSTALLED" | sort | column -t echo "" } gpI () # list installed packages { INSTALLED="$(PACKAGES="$(guix package -I | grep -v ^font-)" echo "$PACKAGES" | while IFS= read -r PACKAGE; do NAME="$(echo "$PACKAGE" | cut -f 1)" VERSION="$(echo "$PACKAGE" | cut -f 2)" OUT=":$(echo "$PACKAGE" | cut -f 3)" if [ "$OUT" == ":out" ]; then OUT="" fi echo "$NAME""$OUT" "$VERSION" done)" echo "" echo "$INSTALLED" | sort | column -t echo "" echo "$(guix package -I | wc -l) packages in profile" echo "$(guix package -I | grep ^font- | wc -l) font packages" echo "" }

If i'm a windows user tired of bullsoykaf, living in the third world, in a 'dictadura cibernetica latinoamericana ', this is the OS to me? sorry for redaction, i'm working class latino trying the goverment don't kill me. Saludos

>>16963 In a computationally centric society, when you don't have control over your computing, whoever does has control over you. Free software becomes paramount to freedom at all in general. It is not foolproof and alone will not save you, but removing proprietary software from your life is the most basic necessary starting point. I wish something like GuixSD existed when I was first making the move over - I think it would have made everything a lot more clear and easy. But that's just me, I suppose others might find it totally inaccessible and eclectic... If you can follow directions, give it a try and let us know how it goes.

>>16964 Thanks, i'm here to learn, things here are dangerous if you don`t like 'state of control', problems come if they find something, but finding nothing in your technological environment you don't die, thanks again.

>>16959 That's beautiful, I'm extremely interested. Quick question, is it possible to do source based management in Guix like in Nix and Portage among others? I'd love to switch to a system like this but also quite like the ability to compile my programs with debug flags linked to source on the file-system.

I'd like to install Guix System on my Rock64 board, but there's no system for that. #:export (installation-os a20-olinuxino-lime-installation-os a20-olinuxino-lime2-emmc-installation-os a20-olinuxino-micro-installation-os bananapi-m2-ultra-installation-os beaglebone-black-installation-os mx6cuboxi-installation-os nintendo-nes-classic-edition-installation-os novena-installation-os pine64-plus-installation-os pinebook-installation-os rk3399-puma-installation-os wandboard-installation-os os-with-u-boot)) :sad face:

>>16956 >>16959 >>16962 Hey, if you're just pasting your posts from 4/g, can I also bitch about being unable to boot from the iso here? Jokes aside, I will have to try this again this week-end, in a vm if not on bare steel. Thank you for the examples, I always found guix to be fascinating but a bit hard to approach.

Does it have LUKS yet?

>>16972 I know that it does not yet have Gentoo style use flags or whatever - Is this what you're asking about? >>16974 I'ma try putting this on my BeagleBone Black - Finally give me a reason to do something with it. >>16977 What booting problems? I've never understood what others find hard about it. I've no idea how anything works but have found GuixSD to be the first OS I can finally actually sort of wrap my head around and manipulate meaningfully. >>16982 Yes. https://www.gnu.org/software/guix/manual/en/html_node/Mapped-Devices.html

>>16983 >I know that it does not yet have Gentoo style use flags or whatever Is that even a goal? I thought the point was reproducible builds and you can't have reproducible builds if everyone is using different USE flags.

>>16983 >I know that it does not yet have Gentoo style use flags or whatever - Is this >what you're asking about? USE flags aren't very important to me, I just want to have packages which are compiled with -Og -ggdb in GCC for hacking and debugging purposes. I think Guix lacks this functionality unless you want to run a private repository, which I wouldn't be apposed to. Perhaps I should just move on though, it's quite rarely that I take advantage of these features and it might be that the gains from Guix would be more substantial.

>>16986 > I just want to have packages which are >compiled with -Og -ggdb in GCC for hacking and debugging purposes Sounds like you want systemtap or dtrace. Compiling your whole system with -g is kind of dumb to be honest.

Important Note: Due to a bug, the graphical installer of Guix 1.0.0 may produce a system configuration where essential commands such as ls or grep are missing; as a side effect, it also prevents logging in into Xfce. A new release will be published in the coming weeks. Here is how you can work around it in the meantime on your freshly installed system: Install packages that provide those commands, along with the text editor of your choice (for example, emacs or vim) guix install coreutils findutils grep procps sed emacs vim Change the packages field to add the “base packages” to the list of globally-installed packages, such that your configuration looks like this: (operating-system ;; ... snip ... (packages (append (list (specification->package "nss-certs")) %base-packages)) ;; ... snip ... ) Reconfigure the system so that your new configuration is in effect: guix pull && sudo guix system reconfigure /etc/config.scm

>>16956 The link is useful for everyone else too. >>16972 https://ambrevar.xyz/guix-advance/

>>16956 is the gnu kernel still total soykaf?

>>17135 pretty much

>>17135 are we still pretending it's gonna happen? :D

>>17140 yes

>>17141 hurd must be in development hell at this point, i wouldn't hold my breath. but it's nice gnu is still thinking big.

>>17134 oh man, this guy's site is great

>>17143 agree

There needs to be a bit more documentation before dumbos like me can use this I feel like. I will try to install it in a VM and figure out how the packaging and configurations work. The features look really sexy though.

>>17152 >more documentation before I swear this is some kind of weird Stockholm Syndrome with people coming from Debian/Gentoo/Arch... If you just want a normal desktop, it's the easiest thing to set up in the world - You're problem maybe is that you think you should have to do more or something confusing, which none of it is: >boot to usb >bring up network >partition disk and make file systems >copy or write a few basic lisp declarations ala >>16959 >run a single short command ($ guix system init .../config.scm /mnt/) >reboot into new operating system This is all shown in greater specific detail https://www.gnu.org/software/guix/manual/en/html_node/Manual-Installation.html The manual is extremely fleshed out I feel - Everything you need to know is in there somewhere very clearly explained https://www.gnu.org/software/guix/manual/en/html_node/index.html It's maybe lacking in diversity of examples - But 1) that's not really what a manual should be full of but there's enough that if you know how to read it you should be able to figure out how to do anything else, and 2) you can easily obtain examples of anything looking at peoples gits or asking on irc. Here's a nice tutorial on guix packaging https://ambrevar.xyz/guix-packaging/ but if you just want a nice simple desktop internet machine, you don't need to learn any of this. It's plug and play.

>>16982 >>16983 Ah, I see. I thought LVM was needed for LUKS. What is LVM used for anyway?

>>17153 >plug and play The graphical installer just left me in a box without installing a terminal-emulator or drun on an i3 installation very user friendly and easy :-) Show me where in the manual I can read to setup wayland for example. The manual is just that, a manual. it's not documenting how to use the advanced features in a real world setting? There is 20 lines total about the container option for guix. Have you seen arch/debian/ubuntu wikis?

>>17154 >Ah, I see. I thought LVM was needed for LUKS. What is LVM used for anyway? LVM is often used in combination with LUKS to unlock multiple encrypted partitions at the same time. For example, it's useful on laptops where you want swap space to be encrypted, but don't want to save the LUKS key to disk (potential security risk), and you don't want to type multiple passphrases on bootup (inconvenient).

>>17154 The Linux Volume Manager is an abstraction layer that mapps virtua volumes into physical ones. This allows for multiple drive partitions, snapshots and neet stuff like that.

So I tried 1.0 last night. First the installer wound up not detecting my wifi, which I know linux-libre supports. So after fiddling with the wpa_supplicant.conf and getting it connected manually the installer then booted me into a gnome shell without any shell tools (no cp, ls, man, mv, etc.) and I also found I couldn't even connect to the internet through the usual gnome internet dialog. It just kept saying password not valid, even though I was looking right at it. Maybe these issues will be fixed in a new release.

On the differences between Nix and Guix: https://news.ycombinator.com/item?id=19809366

Has anyone tried this on a x220? What kinda problems should I expect? On my first try it failed to install a bootloader and I left it at that. Kinda curious on giving it a new try.

>>17161 See >>17117 - there's a bug in the graphical installer which meant that %base-packages wasn't added to the list of packages. A bit of an awkward an elementary bug - the dev team has apologised and said it'll be fixed in a later release, and they've also produced a fix. The wifi stuff is funny - maybe see if you can bring it up on the mailing list or in the IRC to see if it's a known issue or something to look into.

>>17189 I'm on my way to having coreboot and guix on my x220. Just figured out how to go about bootloader in guix to have whole disk encrypted. Not running the thing yet, but planning the big leap in couple days.

>>17161 To get wifi working on my system in the installer I had to run 'rfkill unblock wifi' but it sounds to me like you're missing a kernel module for your wireless card, I think by defualt only ath9k and one other are included. Also i had to add nomodeset to boot into the installer if you end up with graphical issues early on.

>>17196 I did that. Steps I took. 1.boot into graphical installer 2.do a full setup using a single partition with GNOME 3.installer claims it can't detect wifi 4.go into shell and configure wpa_supplicant following the docs in the manual. 5.run rfkill unblock all and dhclient -v wlp2s0 to bring up wifi. 6.ping gnu.org success 7.re-run installer, except it STILL can't find my wifi card but continue anyway. 8.it gives a default config.scm and succesfully builds a desktop, but without base packages as detailed in the install manual. 9.Attempt to connect to internet via gnome dialog fails. 10. Unknown fix so give up and go back to Trisquel.

This looks great conceptually, but it's riddled with GNU so it's probably gonna stay on server.

>>17194 Alright, going to try it on my librebooted x200 and hope that it works without any hassle.

Looks like they released the 1.0.1 bugfix already! Gonna try again and hope it goes better than last time. https://www.gnu.org/software/guix/blog/2019/gnu-guix-1.0.1-released/

>>17216 You could try Nix if the whole GNU stuff seems off to you. Been using and contributing to Guix for quite some time and it has been a joy. It has replaced all package managers for me and I use to deploy rails, python and other apps. Even managing my emacs packages with just guix. It feels quite liberating, though there are some problems sometimes. The fact I can build a system and easily reproduce it in various machines is really handy.

I remember giving GNU GuixSD a go before it was released as 1.0, it unfortunately wasn't as stable as NixOS at the time. But with the 1.0 release I see things have changed a bit. Checking the GuixSD documentation it seems GuixSD would largely be able to replicate the environment I have running on NixOS right now . But one thing that I cannot figure out is how I'd go about handling Suckless tools with their compiled config.def.h files (or similar). In NixOS one way of solving it is a package override: nixpkgs.config.packageOverrides = pkgs: rec { st = pkgs.st.override { conf = builtins.readFile ./st.h; }; dwm = pkgs.dwm.override { patches = [ ./config.def.h.diff ]; }; }; How would I go about achieving the same thing on GuixSD? The best I was able to find was a reddit post (/r/GUIX/comments/bg6y6s/guix_and_stdwm_configh/), which didn't really help. On a related note I just can't help but be reminded how much better the Gentoo/Arch documentations are. I'm sure the Guix docs are great for experienced users but I'd claim they're not too approachable for a beginner, Gentoo/Arch do a better job here. NixOS documentation suffers from this too. I now feel the same as I did when I was learning to use NixOS largely by looking at other peoples configuration.nix files when the official wiki was insufficient for what I was trying to do.

The workaround for >>17206 in my case was to use a cable from the router connecting to the laptop. Working good now, I suppose.

The repo is missing a few essential packages I made use of regularly like a good email client like Thunderbird or icedove. Seriously, that program makes things alot easier and convenient for me. Luckily I got help with configuring gnus in Emacs a little, but it's not perfect and I wish an easier solution existed already. Luckily new packages are being made available regularly, so it might not be too long before then

>>17368 You could always help by packaging icedove, or whatever. It should be pretty easy and you can even import package definitions from nix if something is packaged there. Generally it was pretty easy for me to push upstream packages that I need, and what was not complying with the ideology I put in a separate personal channel.

Is there a way to install nonfree firmware? I want to try it out (which i will anyway) but i have a t430 and i want to get wifi working without changing cards or whatever

>>17873 Yes you can roll your own kernel, although that is not officially supported. Check here: https://gist.github.com/TeMPOraL/306bbb7c627ab0d32fa84af831963563 and here: https://www.gnu.org/software/guix/blog/2019/creating-and-using-a-custom-linux-kernel-on-guix-system/ If these seem too much to start with, use your current distro and just use guix as a package manager. You more than 70% of the experience.

>>17873 Second to what the other poster said: check out https://gitlab.com/nonguix/nonguix for more up to date package definitions. Also, check out https://github.com/ison5002/guix-g for some help on using nonfree Linux as your kernel in Guix. This repo also has a fairly up to date kernel, it’s just also got a bit of ... /g/-related fluff. Still very useful though.

Trying guix right know in a vm and liking it a lot. But two questions: 1. If you use LUKS encryption you get two password prompts: one for /boot and one for root. Is there a way to not encrypt boot partition? Two prompts are really annoying, 2. Maybe someone knows of a repo with qtwebengine package? And I know, chromium bad and all, but large chunk of qt programs won't work without it and you get a bunch of CVEs...

Tried to change gdm to sddm for wayland support, didn't work, reverted to 1st generation and was greeted with this. Damn, staying on Ubuntu for now I guess.

I jammed a small ssd in my thinkpad and installed Guix System on it, works pretty good so far and installation went flawlessly. It takes some getting used to but it's a good excuse to finally get in depth with lisp. Pretty fun overall.

>>17881 >check out https://gitlab.com/nonguix/nonguix I followed the steps but I keep getting "no code for module (nongnu packages linux)"

>>18132 Nevermind found it.

I installed guix system on a spare laptop, and I rather like it so far, the installer is very easy to use. But you really have to learn Guile before trying to modify the config.scm and reconfigure. I lost hours trying to wing it and solved my issues right away once I learned the basics of Guile. I hope to have Guix on my main laptop soon enough.

One question that isn't clearly answered in the doc (as far as I understood it) : Can the various dot files be specified in the config.scm? For example if I want a specific .bashrc or .gitconfig for each user, do I have to copy them after I installed the system, or can it be included in the config.scm? I imagine that you could just some guile function to call a bash script that would do that. But maybe there is a cleaner, more specific function for that?

>>18147 For user-configs -- the ones that usually reside in ~ -- I don't think there is an abstraction to create them. If you need global confs -- think of configurations in /etc -- then almost every service has some configuration abstraction which you can define in your system config and then create the appropriate config files. Of course this doesn't cover every use case, but it's useful for fine-tuning your system.

Would you guys consider having nothing but free software int the main package repository a hindrance? a challenge maybe? I know that the whole point of using GNU software is freedom but there are some things you can't really avoid when it comes to software freedom, like wifi drivers for example

>>18158 Seeing how easy it is to package and make channels, I don't think so. Getting the nonfree soykaf the work is pretty basic but in the best case it will guide some "open-source folks" towards the idea of free software. It might suck getting assistance when something nonfree is being a bitch since asking about it is mostly a big no.

Does it support LVM yet? Encryption is annoying without it and I have a lot of LVM disks I'd like to mount.

>>18169 Nope

I am really interested in guix, but I have a couple of questions: first, what would be the main advantage of installing the whole guix system over just the package manager in my current distro? second, is the whole operating configurable through guile? seems like the answer is yes, but I want confirmation. I will probably install it on my secondary laptop before i jump it to my main machine. the main reason, i do some light gaming now and then. the ambrevar blogpost says its great for gaming (O_o) but does anyone know how is steam and nvidia drivers support?

I have this piece of GNU-licensed software: https://github.com/xtrx-sdr/xtrx_linux_pcie_drv/blob/ab29a2b4a319ae225fe38bffc185545c997b4e85/xtrx.c And I can't use it. GuixSD seems to have no dkms, but that's ok, I can build my own .ko. But linux-libre-headers package seems to miss some of the headers necessary to build this source, and there's no /lib/modules/build environment to build a kernel module. What do?

>>18306 >first, what would be the main advantage of installing the whole guix system over just the package manager in my current distro? The biggest reason is the fact that you get transactional package management over your entire operating system. In effect, every OS configuration description is its own distribution of GNU, and you can get all the benefits of reproducable builds and transactional operations over that distribution. You get the benefit of generations, which allow you to roll back your operating system definition to a previous state. The implications of all this are wide and very interesting. You can even version control your operating system - just stick your OS configuration file in a git repository. >second, is the whole operating configurable through guile? seems like the answer is yes, but I want confirmation. Essentially, yes. You might end up with some plain text configuration in your operating system definition for some services, but yes, the whole operating system is configurable through guile. >the ambrevar blogpost says its great for gaming (O_o) but does anyone know how is steam and nvidia drivers support? Can't really speak on this because I'm not really a gamer, but at least in theory it shouldn't be any worse than any other distribution of GNU with linux-libre. There are a number of Guix package definition repositories out there which will give you standard linux (particularly useful for non-free wireless drivers), and I'm pretty sure I've seen some graphics drivers in there too.

I'd like to migrate to guix, however I'm not familiar with guile (or any other lisp language) yet, should i learn it first, or just pick it up along the way?

Guix 1.0... did they finish the GNU kernel slash Hurd yet?

>>16958 how could they possibly roll a GNOME desktop without having systemd?

>>18489 Gentoo does that too.

>>18489 there is elogind that is standalone version of systemd's logind. check https://github.com/elogind/elogind if you are interested for more. in general it works well, i think the main problem is that you can't set timezones through system settings. i'm using gnome since almost its first appearance in guix and it works just fine. we're still waiting for 3.30 though, i think it will get merged soon.

>18482 As someone who can't grasp code, I thought it would be a bit easy to switch to Guix. However the docs are lacking, I feel for novice users to grasp the system configuration. I would like to learn Guile, but I have no idea where to start.

>>18499 Please go on and to install it and ask anything here. I feel this article is a good introduction. --> https://ambrevar.xyz/guix-advance/index.html It is true that the docs assume quite a lot of prior knowledge for the user. There are some videos in the making that are considered to be more noob friendly. I'm checking the board quite often so I'll try to help. Also check #guix on freenode, there is almost always traffic and people are quite happy to help.

>>17153 *on backdoored proprietary x64 If however, you're on an ARM board with open source firmware, none of this works for you

>>18541 What I am trying to do so far is to just get a proxy service started. I want an Openvpn configuration file to be run without root privilege on startup and then be able to start for after. Can I make a script or something for this?

>18631 Tor

>>17134 >systems like Arch Linux and Gentoo: because only a few versions are supported at a time (mostly just one) But that's bullshit for Gentoo. >Guix channels advantageously replace Arch Linux’ AUR and Gentoo’s overlays, allowing anyone to distribute their package definitions from, for instance, Git repositories. Which is exactly what an overlay is. >None of the changes have to be carved into the package definition. At any time, the user retains full control over the changes that can be made to the packages. USE flags aren't set in the ebuild. >For instance, if a package has optional audio but the packager didn’t expose the corresponding USE flag, the user cannot do anything about it (beside creating a new package definition). Which is the "right way" to do it, since Portage must keep track of what USE flags was enabled to allow binary distribution to work (i.e. make sure someone asking for a binary package has matching USE flags; no surprises here). Also, this nullifies what was said before: how can you loop over packages and enable audio (for example) if the build system doesn't have the same syntax? USE flags are primarly that, a way to map autoconf/cmake/meson/... options to the same name and a way for packages to depends on package with a feature list added. >Guix completely nullifies this complexity layer by using Guile Scheme to program the relationships between packages. I don't see how using a different language means that the complexity is gone. I prefer Scheme to Python, though; but I prefer Tcl even more. >Guix provides pre-compiled packages which can be a huge time-saver compared to Gentoo `eix -c '*-bin' | wc -l` gives 99. Only for the biggest and meanest packages (qtwebengine doesn't have one, sadly). >Gentoo famously “requires” a custom kernel as a recommended (mandatory?) installation step. This is hardly declarative however, and the users have to maintain the kernel’s configuration themselves. Genkernel or just download Debian's config. Honestly, I do think that Guix is very nice (that Coc, though) but I don't think that guy knows how Gentoo works, seeing the elementary mistakes he makes.

>>17134 Also >Pike was wrong. And this proves another more general point: it’s probably wiser to refrain from claiming that any research has become irrelevant, unless you can prove that no further development is possible. Does he realize that piling more and more soykaf in the userland to try to fix the whole POSIX mess can't be considered developement? Especially compared to the beauty of Plan 9.

Would there be any value for a NixOs user to switch to guix? Would I loose my comfy Nix dev environment?

>>18644 >Would there be any value for a NixOs user to switch to guix? Would I loose my comfy Nix dev environment? The main differences between GuixSD and NixOS are: - Guix doesn't badly reinvent a programming language used to extend the system - it uses Guile Scheme instead. Thanks to this you can use libraries and modules already present in the language, giving you more power over your configuration. - NixOS uses systemd, which is really bloated and hard to understand, wheres GuixSD uses the GNU Shepherd - a process supervisor made entirely in Scheme and the configuration is in Scheme too. Here's also visible another feature of Guile - besides being a programming language it is also an extension language - you can make your C, or Scheme program configurable in Scheme, don't know if the language Nix uses allows that, or are there always wrappers and translators? But I havent seriously used Guix nor Nix as a daily driver yet, because I don't have a libre computer yet.

>>18649 Nix has the ability to act as a wrapper around patches for something like st.

>>18650 >Nix has the ability to act as a wrapper around patches for something like st. I have no clue what 'st' is, please explain. Guile provides libguile - you can include it into your program and use Scheme from C, this allows you to decide what do you want to be accessible from Scheme interface - the same way Emacs uses elisp, if you're familiar with it. Can you do the same with the language Nix uses? https://www.gnu.org/software/guile/

>>18651 st [1] is a terminal emulator developed by suckless [2], famous for their minimalist ideologies and programs. As with most of their programs, a lot of features for st are implemented by user patches (for instance scrollback, colourschemes etc). st is also one of the smallest terminal emulators which supports true color (unlike urxvt) and other stuff out of the box. Personally, the suckless ideology can be a bit of a stupid thing sometimes (they use SLOC as a metric for good code). I do however like using st and dmenu (another one of their creations). References: [1]: https://st.suckless.org [2]: https://suckless.org

>>18669 Using SLOC is stupid but using order of magnitudes of SLOC isn't.

>>18650 I've just searched Nix's documentation a bit and I couldn't find any information about using Nix expressions as an extension language in programs like with Guile. https://nixos.org/nixos/nix-pills/basics-of-language.html >The Nix language is used to write expressions that produce derivations. It seems Nix language is just a badly designed domain-specific language and you can't do much more with it. Guix does this one better. But I didn't check everything, if it is possible to use Nix language as an extension language, please tell me. >>18669 >st [1] is a terminal emulator developed by suckless Ah, this thing.

Would any of you fine fellows be interested in contributing to an imageboard written in Guile, heh?

>>18689 >Would any of you fine fellows be interested in contributing to an imageboard written in Guile, heh? I'm currently learning Guile, but I hate web development.

>>18689 I would, pls provide more info. Are you using Artanis or something? I'm not the most proficient but I have done web programming in Clojure and Common Lisp.

Anyone knows how to run a bash script on Guix System? Just tried running a script with #!/bin/bash and it didn't work. Is there a different way than to change it to #!/gnu/store/hash-bash-version/bin/bash? I mean changing it like that in every script isn't very convenient, is there a better way?

>>18693 Indeed. I left my email so mail me pls.

>>18709 It's .com; it always cuts out in the email field here.

>>18702 I use #!/usr/bin/env bash and have create a symlink from /run/current-system/profile/bin/env to /usr/bin/env. This way I can run any scripts, python, node, sh or whatever. >>18709 I just sent you an email!

>>18736 >I use #!/usr/bin/env bash and have create a symlink from /run/current-system/profile/bin/env to /usr/bin/env. Thanks, will check it out.

>>18736 >I use #!/usr/bin/env bash and have create a symlink from /run/current-system/profile/bin/env to /usr/bin/env. Do you create the symlink manually? Because I've found out you can add a special file using service configuration in config.scm: https://guix.gnu.org/manual/en/html_node/Base-Services.html#Base-Services >If you want to add, say, /usr/bin/env to your system, you can change it to: scheme `(("/bin/sh" ,(file-append bash "/bin/sh")) ("/usr/bin/env" ,(file-append coreutils "/bin/env"))) >Since this is part of %base-services, you can use modify-services to customize the set of special files (see modify-services). But the simple way to add a special file is via the extra-special-file procedure (see below.) >Scheme Procedure: extra-special-file file target >Use target as the “special file” file. >For example, adding the following lines to the services field of your operating system declaration leads to a /usr/bin/env symlink: scheme (extra-special-file "/usr/bin/env" (file-append coreutils "/bin/env"))

>>18784 That's exactly how I do it. I don't like manual with guix, I would also like to handle my $HOME configs like this, there was some discussion in the mailing list, but for now it seems too complicated.

Will my gosh damn laptop wifi work on it? I swear, it's the only thing holdin me back from goin full GNU. This looks promising

>>18797 https://gitlab.com/nonguix/nonguix as said before. You can manage your own kernel, or get a dongle.

>>18797 >Will my gosh damn laptop wifi work on it? What model? Check the list of hardware being able to run free software: https://h-node.org/ I've recently bought a librebootable Thinkpad to run Guix System, still I have to buy a wifi adapter that'll work with Linux-libre. Also you can install blobs, like >>18799 said.

I installed Guix a few months back, and I noticed that when I install a new tool, it doesn't appear in my DE's menu. Was this a mistake on my part, or a structural fault?

>>18820 >I installed Guix a few months back, and I noticed that when I install a new tool, it doesn't appear in my DE's menu. Was this a mistake on my part, or a structural fault? Installed Guix a few days ago and I have the same problem. Notice when installing a package, Guix writes "XDG desktop file cache..." and "XDG Mime database...", XDG is a set of standards describing common things across many DEs, including desktop entries. I suppose certain DEs are not smart enough to handle what Guix does, but idk, maybe someone should report this as a bug. I use Xfce.

Hello Lains, I've been using Guix System for about month and although you can encounter some bugs, if doing nonstandard things, the system looks very promising. I really like the idea of defining your system with a config file - you can try many DEs, Wms without breaking things. The "guix pack" option is also a convenient way to share packages in a reproducible manner, across distributions which don't yet have Guix - it's like flatpak or appimage, but better. Is there any package manager allowing you installing packages on machines without the package manager or is Guix alone? Guix News: Guix has been successfully built on Debian GNU/Hurd, now developers are solving problems with building bootstrap binaries. This will eventually provide a GNU/Hurd port of Guix. Hope the Hurd will gain some new developers, after making a usable GNU/Hurd distribution. Discussion on the mailing list: https://lists.gnu.org/archive/html/guix-devel/2019-09/msg00000.html GNU Mes 0.20 has been released, bringing us closer to the vision of bootstrapping the OS entirely from source code and a human-readable assembly file (stage0). Interesting part is that the next target to bootstrap is the Hurd: >Next targets: > - ARM support > - Reduced Binary Seed bootstrap for ARM > - Scheme-only bootstrap: use Guile and Gash to remove bash, > coreutils&co, grep, sed, etc. from the Guix bootstrap binaries > - mes-m2: port Mes.c to M2-Planet > - Introduce Reduced Binaries Seed bootstrap to NixOS > - Debian? > - Hurd https://lists.gnu.org/archive/html/info-gnu/2019-09/msg00002.html So in the relatively near future we can see a fully bootstrapped Guix System on GNU/Hurd

>>19949 >Guix has been successfully built on Debian GNU/Hurd, now developers are solving problems with building bootstrap binaries. This will eventually provide a GNU/Hurd port of Guix. Hope the Hurd will gain some new developers, after making a usable GNU/Hurd distribution. Realisitically speaking though, what would the real benefit be of using Hurd?

>>19998 >Realisitically speaking though, what would the real benefit be of using Hurd? The microkernel based design is better than monolithic design of Linux because: - it is more stable - on Linux if a driver or module crashes, the whole kernel crashes. On the Hurd if a subhurd crashes, nothing really bad happens - other subservers can turn the faulty subserver back on. - it is easier to maintain and develop - object oriented architecture allows adding new features, without compromising the design. If you want to add a driver to Linux you have to "git clone" the whole Linux repo - about 2GB and you can't develop your driver outside of the source tree, because the API of Linux is rapidly changing. On Hurd you just create a subserver without carying much about the mikrokernel - GNU Mach handles only the most important stuff, whereas everything but simple hardware operations is handled in the userspace. - the Hurd gives you more freedom and stability - for example you can develop a new filesystem as a subhurd and test it without breaking the rest of the system or rebooting - you just start and stop the subserver. This means users can add and remove their own drivers without breaking the system. The microkernel design is just better, even Google decided to switch from Android and ChromeOS using the Linux kernel to Fuchsia running on Zircon microkernel, because maintaining Linux is just too expensive. Microsoft's NT kernel was a properly designed microkernel too, but microsoft broke the good work of developers by merging the GUI into the kernel as Win32 subsystem. Despite the fact NT is maintained by incompetent people and the system is buggy, the kernel itself is great - it is easily extensive by subsystems, like POSIX subsystem or WSL (should be actually Windows Subsystem for GNU/Linux) - the entire GNU/Linux system is running inside Windows thanks to NT's architecture. Here you can read more about advantages of the Hurd: https://www.gnu.org/software/hurd/advantages.html https://www.gnu.org/software/hurd/faq/still_useful.html The Hurd is as flexible as NT or even better - you can write a translator subserver translating Linux drivers to rump kernel drivers. A rump kernel is currently being developed for the Hurd. http://rumpkernel.org/ https://git.savannah.gnu.org/cgit/hurd/incubator.git/log/?h=rump In my opinion the Hurd is worth developing and it's really sad people forgot about it.

>>20008 >The microkernel design is just better, even Google decided to switch from Android and ChromeOS using the Linux kernel to Fuchsia running on Zircon microkernel, because maintaining Linux is just too expensive. Google may have some valid technical reasons for a microkernel switch, but I disagree with the "saving money" thing. Google has more money than God. No, I'm fairly sure that Fuchsia is a play at owning the entire Android-successor stack, top to bottom, Apple style. No more AOSP, no more having to release source externally. Just look at the way they've been gradually killing off AOSP apps over the years in favour of Google-account-centric closed-source apps. They'll do the Microsoft Win32 thing where anyone can easily develop *for* the OS with public APIs, but they'll keep the OS itself on lockdown.

>>20074 >Google may have some valid technical reasons for a microkernel switch, but I disagree with the "saving money" thing. Google has more money than God. You get rich by not spending money on things you don't need to. Surely they have a lot of money, but they can have even more if they switch from maintaining a monolithic kernel to a microkernel. Microsoft didn't switch from DOS to NT just for fun, Google isn't switching to Fuchsia just for fun either. Even Apple has a microkernel-based design. They can maintain just one system and one kernel, instead of two systems and two different forks of the mainline Linux kernel. No one wants to maintain a traditional, monolithic UNIX-like kernel anymore - if the quality of the code has to be higher in order to maintain stability, the development is much more expensive. Linux was supposed to be a clone of MINIX, which has a microkernel, but somehow Torvalds made it monolithic. It's a historical mistake, which is popular mainly because it is free software and works with GNU. >No more AOSP, no more having to release source externally. The source code is available in the public repository https://fuchsia.googlesource.com/fuchsia/ But the license is non-copyleft, so the version shipped to users will be proprietary and locked down. I agree with what you said about other reasons for switching kernel, but anyway Google is a corporation and their every decision is always supposed to be profitable. Even if saving money on development of the system(s) isn't the main reason for Google switching, I think it is one of the reasons. Sage for offtopic.

>>20078 >but anyway Google is a corporation and their every decision is always supposed to be profitable. No, not really. First, a business - if expected to maintain as an ongoing concern - ought to be profitable *overall*. It's perfectly acceptable to have losses in individual lines of business so long as they contribute to overall profitability. That's why it's accepted accounting practice, even in ridiculously huge companies, to lump all expenses and revenues together on income statements. Investors don't care if some support division was a loss-leader, so long as the company itself is showing profitability trends. For example, take a look at Boeing's fiscal-2018 income statement. It's a truly massive company, involved in everything from jetliners to biofuel R&D to military contracts to building NASA's heavy-lift SLS rocket. And yet their income statement (not including headers) is 21 lines long. Second... have you been paying any attention to anything that Google has done in the past decade? It's complete chaos. Decisions are being made all the time by middle management without any thought given to profitability or company-wide plans. The only reason there hasn't been a company-wide come-to-jesus "What the fuarrrk are you doing?" moment on behalf of senior management is because their ad business is so ridiculously profitable that they can afford not to give a damn about profitability of other business lines. (If Google's ad business starts to tank, you're going to see an unprecedented cost-cutting bloodbath. That is one fat company.)

is kde going to be available?

>>20008 >object oriented architecture

>>21094 >is kde going to be available? Eventually someone is going to make it available, but for now I don't see much work. Here's something related to KDE, don't know if it's relevant though: https://lists.gnu.org/archive/html/guix-devel/2019-09/msg00237.html

>>21094 No idea, I hear KDE is dead though. Dev exodus or somethin

>>21165 I mean this has nothing to do with Guix. CMIIW

I'm sorry, but I cannot take guix seriously anymore The lead developer has been publicly attacking Richard Stallman based on mischaracterizations and lies He's trying to kick Richard Stallman out of the clubs Richard Stallman started He and his enemies of free software posted their joint statement on the guix.gnu.org subdomain, then he proceeded to soykaf up the mailing list with his politics. https://lists.gnu.org/archive/html/guix-devel/2019-10/msg00076.html >Not surprisingly, I agree with all the goals you propose for the FSF, >except perhaps one: to not ‘mix our ideas with general politics’. I >think free software is a social movement that doesn’t exist in a vacuum, >it’s politic in nature, and thus it’s part of ‘general politics’. >Ludo’. What an absolute soykaf bag. He wants to have general politics on a software development mailing list, and then ban anyone who doesn't agree with him. To review, Ludovic starts the political flamebait thread, insists that general politics is on topic after RMS says it is not, waits for people to disagree with politics loudly, then starts banning them. https://lists.gnu.org/archive/html/guix-devel/2019-10/msg00178.html >You’ve made your point now; I see you’ve even set up a web page to >collect hatred messages against me. >I ask you to stop using the Guix mailing lists for this now. I will >propose to the Guix maintainers to put you on a moderation queue if you >don’t stop by yourself. >Disagreement is fine; attrition is now. >My sincere apologies to all the Guix people who are witnessing this. :-(

>>21180 I don't like the leadership either, but I do what I think that the anti-RMS gang should have considered too, and accept it. Better than making the pro-RMS position the "crazy" one.

>>21180 >I'm sorry, but I cannot take guix seriously anymore Can we talk here only about technical details of the software please? Let's keep the discussion about RMS in the thread about the case. I don't like what the devs did, but I like their software and will use it.

>>21185 This is correct. While I don't like the recent RMS debacle either, this thread concerns techincal aspects of guix and help. Take the guix politics to the rms general thread.

>>21185 >I don't like what the devs did, but I like their software and will use it. The risk is that if the project starts driving away talented developers because they have the Wrong™️ opinions about politics then the technical merits of the software is not going to last.

>>21275 You're probably the same guy who made that point in the RMS-general, but the problem is that most people who contribute a lot and good stuff aren't really one one "side" or another. Most people just accept whatever rules they have to, agree with "right opinions", and do their job. So now, please, stop spamming this thread.

>>21277 >the problem is that most people who contribute a lot and good stuff aren't really one one "side" or another. Correct. >Most people just accept whatever rules they have to, agree with "right opinions", and do their job. Incorrect. The problem with extremist politics (from any side) is that everyone who is not extreme enough gets soykaf on exactly the same way as if they are on the enemy side. This is especially a problem when the "progressive" crowd are in charge because their modus operand is to constantly move the goal posts of what are acceptable opinions and language and then lynch people who don't keep up quickly enough e.g. "colored person" was the preferred term a few years ago and now makes you literally Hitler. It's interesting you say "job" though. If you mean paid work then you're more or less correct. When I get my corporate "microaggression" training I roll my eyes and nod along like everyone else because I'm paid fuarrrktons of money to do that. But in the open source world there's only so much soykaf you can pile onto volunteers before they FO and contribute to something else. That's not to assert this will definitely happen with Guix. Like I said in >>21275, this is a *risk* we should be thinking about. Since there's no advantage for a technology project to be politically charged in any way, there is a strong argument to stay politically neutral and eradicate the risk completely. >everyone who disagrees with my snowflake opinions is THE SPAMERTROLL Read rule 0.

>>21287 Look, they aren't extreemists. They are at worst corporate pundits, at "best" woke online activists, but that's nothing outside the realm of liberalism. And yes, what you refer to as "extreem" is what most people just accept. What you're doing isn't helping anyone. Unless your "opinion" is that women are worthless, or abolishing slavery is a mistake, you usually will be kindly told that you're insenstive, not nazi-derezzed. Neither RMS by making it seem like this issue is being coopted by the alt-right, nor for this thread, since this *is* a about a technical topic, not your politics. So serve as an example as keep this thread politically neutral, and go back to the general.

>>21290 >my politics is not extreme >only real extremists disagree with me You're only proving my point. Now with the the announcement in >>21180, >I think free software is a social movement that doesn’t exist in a vacuum, >it’s politic in nature, and thus it’s part of ‘general politics’. what does Guix gain from a technical perspective by running people out of the project for expressing nuanced opinions on controversial topics on a private platform (e.g. RMS)?

>>16956 Has anyone had an issue with custom XFCE themes showing red squares instead of checkboxes? It seems to be gtk-3 related, but I can't find out what it is.

This didn't age well.

The system becomes mostly great when learning a bit of guile, otherwise i wouldn't really recommend it, it works but feels a bit clonky. However after just learning the basics of the system configuration, it becomes a blast. Really cool.

Don't know if you saw this, but they reduced bootstrap seed by 50%: https://guix.gnu.org/blog/2019/guix-reduces-bootstrap-seed-by-50/ I can't wait for it being fully bootstrapable from the source code. Guile 3.0 is also going to be released soon, what is going to speed up the system thanks to JIT compilation. There's also an interesting thing happening with the Hurd - for two last months many commits have been merged, some of them include improvements of rump kernel (allowing to run drivers from other systems), SMP support is getting better - some bugs have been resolved. Also one dev got funding from NLNet foundation for adding sound support for the Hurd and what's interesting NLNet encouraged making more proposals for funding for the Hurd. I think there's going to be a new release soon, because the last version was released 3 years ago and the amount of work done is big. I would give you the links to the mailing list, but savannah is down now. I look forward porting Guix to the Hurd. inb4 they backstabbed Stallman. RMS told us not to divide the community and I'm going to follow his advice. RMS supporters leaving the GNU project is the last thing free software needs now. >>22456 >The system becomes mostly great when learning a bit of guile, otherwise i wouldn't really recommend it, it works but feels a bit clonky. I can confirm this. After a few months of learning Guile, I can make own packages and cool system configurations.

>>22456 How did you go about learning Guile? As a non-programmer, I love the idea of Guix but can't understand even basic nested parentheses and where or how many to put in a basic config.

>>22461 I agree with Stallman.

>>22469 >How did you go about learning Guile? As a non-programmer, I love the idea of Guix but can't understand even basic nested parentheses and where or how many to put in a basic config. Not him, but I just read the Guile manual and if I didn't understand something I grabbed the SICP https://mitpress.mit.edu/sites/default/files/sicp/index.html or tyscheme https://ds26gte.github.io/tyscheme/ Tyscheme is good if you had previous experience with programming, SICP holds your hand through learning basics of computer science. And as for parentheses, if you don't know Emacs already, learn basic keyboard shortcuts and install and learn paredit mode. Initially you'll be annoyed by paredit, but it handles ))) for you and lets you reorganize the code easily. There's also Scheme crash course in the Cookbook: https://guix.gnu.org/cookbook/en/html_node/Scheme-tutorials.html#Scheme-tutorials

I used Guix for the most part this year and haven't found it to be as great the more I got into it. It started out really well with ease of packaging, configuration based system, etc but in the end it turned out to be a complex mess. One could argue that I never really went into the depths of it and hacked around the implementation at it's core but that shouldn't be required to provide a decent interface for configuration. Guix's strong point is packages while it's weakness is configuration and general usage. Packages being moved around here and there from module to module makes maintaining a channel rather difficult in the long run as if you're not watching like a hawk you'll `guix pull` and have it fail after compiling for 10 min. I absolutely loved guix and would have it as my main daily driver again once things settle down a bit more though that's sadly going to take a while from the looks of it. I guess my attention should be on taking the nice things and playing around with building similar systems until I eiter can contribute better or create some replacement for myself. >>22469 Read "How to Design Programs", it's a brilliant introduction and doesn't assume you're ready for a full on first term of MIT. After that SICP is much lighter if you decide to expand to it; however you don't really need to as most guix packaging and config is consists of reading what other packages / configs do half the time (even if you know scheme as the back of your hand). Documentation is catching up and there are more blog posts on it now too so jumping in should be a lot easier than before.

>>22461 So much this, I'm going too keep supporting this OS becasue it's just so fuarrrking cool god damn it, it's like gnu/linux done right

Free software is a goal, not an option right now. The amount you're throwing away for bad faith politics is astounding. If you lains really cared about free software you would help make it better instead of using badly programmed, insecure and backwards-looking software to try and own the companies who are destroying society through technology. I dream of a Freely Licensed Micro-Kernel based OS with a software stack that doesn't undermine itself at every single point in insecure and faulty design but that dream will never be a reality as long as companies keep hiring fantastic engineers for awful end products and free software fanatics will wage slave awful incompetently designed software as an alternative. >Owning the proprietary software industry by forcing yourself to run insecure code

>>22544 >free software fanatics will wage slave awful incompetently designed software as an alternative. Could you elaborate the above sentence or show some example for the case? I've compared many free softwares with non-free softwares, but almost every turn, the better one was a free software even in the aspect of the competency. For example, Emacs is the best text editor and it is a free software. (Some people think that vi is better, but I think that is a difference of tastes and moreover, vi is a free software too.)

It's funny how this went from everyone's favorite project to the subject of widespread disgust because of one dev

>>22987 Meh, I still love the project and don't really care enough about the whole drama. If it works it works, it's free too

>>22987 I do still like the features of the OS, but not enough right now to make the switch from Debian or change package managers

>>23007 Why not both and install the Guix daemon on your debian system?

>>23008 I would, but not sure if I want to make the effort again yet. My Guix system install was buggy and not enough packages to suit my needs.

Is the installer still a piece of soykaf? Last time I tried it took a lot of work to even get the ISO to work, then I couldn't get an encrypted install to work no matter what.

>>23019 I had problems on the 1.0 and the 1.0.1 version. Neither can really handle network configuration well. I had to actually just forego the automatic way and do it all by hand. It took awhile since I can't understand how to configure the base system config from the docs.

>>23019 Didn't have any troubles, what version was it when you tried?

>>22461 Guile 3.0 is also going to be released soon, what is going to speed up the system thanks to JIT compilation. https://www.phoronix.com/scan.php?page=news_item&px=GNU-Guile-3.0-Released It finally dropped!

>>23794 >Guile 3.0 is also going to be released soon I was going to post about it, because they're preparing a new Guix release that will be ready in a few weeks, but you were faster. Still waiting for the hurd tho.

>>23796 >New Guix release Neat! I do hope it gets POWER9 support at some point too.

Anyone any experience with running guix in termux?

>>23866 Doesn't seem to be possible yet, but there has been some work on it: https://github.com/termux/termux-packages/issues/3240 https://github.com/t184256/nix-in-termux also looks relevant.

Are there any other interesting things written in Guile beside Guix?

>>24032 >Are there any other interesting things written in Guile beside Guix? The Shepherd - process supervisor, replacement for SysV init https://www.gnu.org/software/shepherd/ Mcron https://www.gnu.org/software/mcron/ Artanis - web framework https://www.gnu.org/software/artanis/ Gash - POSIX-compatible shell https://savannah.nongnu.org/projects/gash And you can check for software written in guile, by checking what's already packaged in Guix: https://guix.gnu.org/packages/G/page/11/ https://guix.gnu.org/packages/G/page/12/ Everything starting with "guile-". If you're looking for something cool to work on, wayland protocol implemented in guile would be cool, at least for me. You could also work on the scheme shell: https://www.gnu.org/software/guile/manual/html_node/The-Scheme-shell-_0028scsh_0029.html#The-Scheme-shell-_0028scsh_0029

>>16956 Does the logo intentionally resemble the female reproductive system?

>>24077 >Does the logo intentionally resemble the female reproductive system? You see what you want to see :) For me it's a head of gnu.

>>24079 >You see what you want to see :) >For me it's a head of gnu. Was this Freudian trap in the logo intentional?

Anyone else having problems running mpd/mpc?

>>24122 >Anyone else having problems running mpd/mpc? I can try installing it, but have no clue how to use it.

>>24123 >>24122 https://wiki.gentoo.org/wiki/MPD is a good place to start with config.

>>24123 >>24129 It's fine, didn't specify the right port Duh

I'm trying this out in a VM right now, will report back

>>24170 Back, and this is suprisingly comfortable. If I wind up liking it enough I might install it to one of my externals and actually use it beyond a VM. Does anyone have recommendations on what to do with it?

>>24174 >Does anyone have recommendations on what to do with it? Well, it's mostly a package manager, so you use it to install software. On Guix System you can explore many different DEs/WMs without the fear something will break (excluding bugs and wrongly packaged software). Just open your /etc/config.scm file and start hacking. Using Guix is a good way to learn Scheme. You can also check out the Cookbook https://guix.gnu.org/cookbook/en/html_node/ and follow the packaging tutorial. You can start from something simple, that isn't yet packaged.

The GNU Hello package was successfully built using Guix on the Hurd. Now they're fixing some other soykaf with it. We can have a ready Guix GNU/Hurd port soon. https://lists.gnu.org/archive/html/guix-devel/2020-03/msg00081.html

https://www.phoronix.com/scan.php?page=news_item&px=GNU-Guix-Going-Hurd HURD GANG HURD GANG HURD GANG

I still can't take this distro seriously after major figures attacked uncle Stallman.

>>26365 another distro commiting hardware suicide how dumb can developers be

>>26365 Cool, it's nice to see Hurd getting some love

>>26380 Obviously an April Fools joke, but there's no reason why it shouldn't be Hurd native if it's Linux-libre native - it's free software after all. Hurd is very neat and cool, it's just not ready for general release (and probably never will be - and that's okay, because we've got Linux, and there's nothing wrong with experimenting with what's a very interesting idea, i.e. a free multiserver microkernel). >>26378 Separate the software from the author. If all you've got against this OS is their blog then you need to grow up. Guix is very happy to be called a GNU system, because the maintainers are responsible enough to understand that just because Stallman's name is attached to GNU doesn't mean that the project is all about him, it's about the community of free software developers. Also Stallman isn't god. He copped it pretty hard from #metoo, but if he'd accepted that he had to look in the mirror it would have been a lot better. Linus did (and rightfully so) and the whole free software community is better for it.

>>26385 Stallman literally did fuarrrking nothing wrong and it turns out his MIT friend was innocent in the end anyway. Not another step back, apologist.

>>26401 If you read the blog post Guix made, they say that he's not a leader and GNU is more than him, and that although his advocacy of free software or his founding and continuing influence of the free software movement can never be denied, he has been and continues to be liability to the movement. Literally all of this is true. The GPL used to be the most significant free software license, now it's almost a historical footnote were it not for the fact that a lot of projects that started last millennium (Linux included) are licensed using it. No new large projects are licensing software using the GPL - they're using Apache and MIT. Why is that? - the GPL is better for everyone involved, even large corporations who get to use the collective labour that has been put into projects like Linux for free, so why aren't more projects using GPL? Stallman has had decades to work with this, and the early 2000s were a fertile ground for capturing the entire world of open source software. He couldn't do it though, because he's an advocate, not a leader. I'm sorry, the leader of the free software movement shouldn't be picking his toenails on stage. Someone like Linus would be better, though he's obviously not interested. Honestly, I think Ludo from Guix would be a fantastic leader of GNU, since he's able to coordinate the work of dozens of major developers on Guix and scores of other small-time developers, and he's demonstrated both his commitment to free software and the fact that he can walk the walk on developing quality software himself (just like Stallman did with Emacs and GCC) - it'd just be a matter of whether his clear ability to manage people can scale.

>>26406 Ludovics is the piece of soykaf who threw Stallman under the bus in the first place, he violated the Guix project's own community rules when he did it, he is a dishonest and inconsistent individual who shouldn't be anywhere near a spokesman of free software.

>>26407 Rightly criticising someone in what is (or at least what should be) a democratic structure is not throwing them under the bus. This kind of sentiment is exactly what is wrong with most of the vehement 'supporters' of Stallman. I say 'supporters' because you're not really supporting him, you're doing him a disservice by being deliberately obtuse and dishonest. For all his faults Stallman has actually been pretty gracious through all of this. Its the zealots who consider change in the free software movement to be a culture war - which it is, but not in the way that you think it is - who are the ones doing damage to the free software movement. Also, Ludo never violated Guix's community rules. Discussion and dissent are allowed within Guix and within GNU, and in reality anyone who is a contributor is allowed to make a blog post. This criticism has been brought up a number of times, and it's pedantic at best - you'll convince no one by making an outlandish interpretation of a project's community rules.

>>26385 >Linus did (and rightfully so) and the whole free software community is better for it. How is it any better for this?

>>26408 Thanks for your thoughts officer.

>>26460 >Thanks for your thoughts officer. This lainon is right though. Disagreeing isn't the same as throwing under the bus. Ludovic has been maintaining a completely free, official GNU distribution for over 10 years already and he together with other GNU developers made efforts to complete GNU as a system by adding support and maintaining the Hurd. Helping finish Stallman's work of life isn't backstabbing. Stallman was the right person to start the free software revolution, but he is not a good leader - a good leader wouldn't let the twitter drama happen. Stallman is also too autistic to work with people and that's why GNU hasn't been completed for over 30 years. The real traitor here isn't Ludovic, but the idiot that leads GNOME. While Ludovic was answering questions regarding Stallman on the mailing list and decided not to drop GNU, this buttsucker Neil whoever told he would drop GNU, if Stallman didn't resign.

THE YEAR OF THE HURD BEGINS!!! https://guix.gnu.org/blog/2020/a-hello-world-virtual-machine-running-the-hurd/ This will speed up development thanks to reproducibility and bootstrapping automation. I think having a GNU/Hurd distribution better than Debian will attract new developers. Now we're waiting for the year of GNU/Hurd desktop.

>>26502 Thanks for make me remember about Neil from Gnome, I had forgotten that. GNU probably shouldn't have given in to such an extortion. This whole cancel culture is a problem.

>>26542 GNOME was stinky even before - they didn't mind putting proprietary software into their "Software" application. >GNU probably shouldn't have given in to such an extortion. What do you mean by this? I think the problem about GNU was being Stallman-centric. The current, more decentralized structure of GNU is better, because you can't just take down the entire project by cancelling just one person. Now it's like a fuarrrking hydra with each contributor being one head.

>>26542 >>26544 Oh, and forgot to mention Neil removed every comment telling it is "abusive" or some other buzzword. I watched the drama on 4ch that time. They just told him that Stallman didn't defend Epstein, but he removed those comments anyway - these were one of the first comments. Meanwhile on guix-devel ML, the only one who got silenced was Jelle Louis (don't remember exactly), because he was constantly spamming the same message. After 3 days of spamming or something like this. Ludovic and other Guix maintainers answered every question people had.

If anyone can help with this, that would be great. I currently have Guix installed on a Devuan install. However, I can only run Guix commands when the daemon is started under root. I cannot figure out a way to have the daemon enabled permanently. Any advice is greatly appreciated.

>>26592 There's this half-assed init script that was posted somewhere. It's not perfect, but works. You put this in the /etc/init.d/ folder, say, named "guix-daemon". Then run "sudo update-rc.d guix-daemon enable" and then "/etc/init.d/guix-daemon start" and it should work. If not, restart your system. #!/bin/bash ### BEGIN INIT INFO # Provides: guix-daemon # Required-Start: $remote_fs $syslog # Required-Stop: $remote_fs $syslog # Default-Start: 2 3 4 5 # Default-Stop: 0 1 6 # Short-Description: Guix build daemon # Description: Provides a daemon that does builds for Guix ### END INIT INFO set -e set -o pipefail mkdir -p "/var/run" case "$1" in start) if [ -f "/var/run/guix-daemon.pid" ] then pid="`cat /var/run/guix-daemon.pid`" if ps www "${pid}" |grep -qa guix-daemon then exit 0 else exit 1 fi else daemonize \ -a \ -e /var/log/guix-daemon-stderr.log \ -o /var/log/guix-daemon-stdout.log \ -E GUIX_LOCPATH=/var/guix/profiles/per-user/root/guix-profile/lib/locale \ -E LC_ALL=en_US.utf8 \ -p /var/run/guix-daemon.pid \ /var/guix/profiles/per-user/root/current-guix/bin/guix-daemon \ --build-users-group=guixbuild fi ;; stop) if [ -f "/var/run/guix-daemon.pid" ] then pid="`cat /var/run/guix-daemon.pid`" if ps www "${pid}" |grep -qa guix-daemon then kill "${pid}" sleep 10 kill -9 "${pid}" || true exit 0 fi else exit 0 fi ;; status) if [ -f "/var/run/guix-daemon.pid" ] then pid="`cat /var/run/guix-daemon.pid`" if ps www "${pid}" |grep -qa guix-daemon then echo "guix-daemon is running" exit 0 else echo "guix-daemon has a stale pid file" exit 1 fi else echo "guix-daemon is not running" exit 3 fi ;; restart) "$0" stop "$0" start ;; force-reload) "$0" stop "$0" start ;; *) echo "Usage: $0 (start|stop|status|restart|force-reload)" exit 3 ;; esac

>>26593 You're a lifesaver, thanks! Second question, is it normal for guix pull to take so long? I understand my T60 isn't the most powerful system, but it's been quite awhile.

>>26546 >Meanwhile on guix-devel ML, the only one who got silenced was Jelle Louis (don't remember exactly), because he was constantly spamming the same message. This is what cemented by belief that the maintainers of Guix are some of the best advocates and examples that the free software movement have at the moment. The restraint and level-headedness they showed even when someone seriously abused a trust system and lead to days of wasted time and effort (this was while the maintainers were trying to coordinate a feature freeze IIRC) is laudable. I recall other developers being warned for basically reacting negatively to Louis, because that's exactly what he wanted. >>26542 Cancel culture is a problem, you're right. But holding people to account and recognising the writing on the wall for people in untenable leadership positions isn't necessarily cancel culture. I don't think Stallman needs or needed to be cancelled, he's obviously a good person whose comments were at best tactless (if not wrong). GNU needs new leadership though, the project is doomed to failure otherwise.

>>26594 (Not OP) It's normal for it to take a long time sometimes, especially if you haven't done a guix pull for a while. How long is 'quite awhile'? - I've had it take hours in a VM, and I recall it taking maybe not quite an hour on my X200. There is a way of making it more verbose if it appears to just be hanging, take a look at the documentation for guix pull if you like.

>>26594 >Second question, is it normal for guix pull to take so long? Pulling itself can take up to 20 minutes, but it depends how outdated you are. >I understand my T60 isn't the most powerful system, but it's been quite awhile. T60 is a toaster. To speed up compilation enter the tty3 by pressing Ctrl+Alt+F3, log in and type sudo herd stop xorg-server if you're on X, then again enter the tty3. Without xorg-server running you'll be able to use the whole CPU by adding "--cores=2" flag like this: guix pull; guix upgrade --cores=2; sudo guix system reconfigure /etc/config.scm --cores=2 After your system reconfigures properly, type sudo reboot

>>26602 Actually, pulling itself can also be fastened guix pull --cores=2 Because if I'm right, pulling is the stage where guix itself is downloaded and package definitions written in Scheme are compiled.

Guix 1.1.0 Released. We are pleased to announce the release of GNU Guix 1.1.0. This release corresponds to 14,078 commits over 11 months by 201 people. It includes many new features, user interface and performance improvements, as well as many bug fixes—see below for a list of changes. Read more about today’s announcement at: https://guix.gnu.org/blog/2020/gnu-guix-1.1.0-released

>>26756 hurray

>>26756 Meanwhile serious Hurd porting effort happens on the wip-hurd-vm branch. https://git.savannah.gnu.org/cgit/guix.git/log/?h=wip-hurd-vm The future has never been so bright for GNU. Just imagine: - reproducible builds - bootstrapping entire distribution from human-readable assembly code - no systemdicks - fully libre - stable microkernel-based system, hotplugging drivers, clean and secure architecture from design

>>26769 And no slave-like CoC in the kernel.

>>26780 >And no slave-like CoC in the kernel. Yet. But to be honest, not being led by megacorporations is more of an advantage here. The modular structure of the kernel lets cheap and independent creation of subhurds/drivers without staying outdated much - something you don't need insane money nor support of Googles and Microsofts. It is a miracle Linux still is usable for desktop, if not Torvalds being sane, it would quickly became a resource hog or server only system. If Torvalds dies, the soykaf will be massive - some optional DRM features are already there and Poettering would surely want kdbus V2 and systemd inside Linux. It is good to have a plan B.

>>26782 Agreed with that. But as far as i can see, software in general and tech related things degrade a lot over time. Or in this case it ether submits to the elites or it turns to trash. More like systemd-kerneld. Also regarding guix; I just gave it a try on my P4, for the second time i think. The new installer still is like kicking dead whales down the beach big time. I tried to use JFS for root partition but it simply restarts without doing anything, it just creates the partitions, despite mkfs.jfs being available. The manual partitioning is still guided: you can't change the layout, only the filesystems and their options. It can't change the default DNS servers in /etc/resolv.conf so i had to do it by hand. Second time i used the default ext4. It spinned the harddisk for a minute then it managed to install a few packages and stopped because "disk is full" after using 200MB from 80GB. Disappointing but i still hope it will get better.

>>26784 >It can't change the default DNS servers in /etc/resolv.conf so i had to do it by hand. Guix is the best once you know Scheme. Here is how the DNS configuration should be done: http://guix.gnu.org/manual/en/html_node/DNS-Services.html#DNS-Services Add your DNS configuration to the /etc/config.scm file, read about it more here: http://guix.gnu.org/manual/en/html_node/System-Configuration.html Installation is like kicking dead whales down the beach, but when you have it done, you just sit and do nothing but "guix pull; guix upgrade; sudo guix system reconfigure /etc/config.scm" for months or years. As for the filesystem, I only tried btrfs.

Would it be possible to run guix (the package manager) on a raspberry pi running raspbian or arch? Last time I tried I got the installation working but guix pull always resulted in errors

>>26806 >Last time I tried I got the installation working but guix pull always resulted in errors Guile had some bugs related to ARMv7 architecture, but they've been fixed already. >Last time I tried I got the installation working but guix pull always resulted in errors Was the guix-daemon up and running?

>>26834 >Was the guix-daemon up and running? hmm, I'm not sure actually. I would assume so because i could install packages with guix.

>>26926 >I would assume so because i could install packages with guix. Then it was running. Build errors can also occur if you're out of RAM memory and the package is bloated, like a web browser or libreoffice. Some packages need up to 8GB of RAM, which isn't the case on raspberry pi.

I'd though to use this as a way to learn a bit of scheme, so I wanted to try and package a simple python package, which seemed easy enough with "guix import pypi" but I keep getting the following error and I'm a bit at a loss PermissionError: [Errno 13] Permission denied: '/homeless-shelter' Does this mean my $HOME isn't configured in the package file?

>>26942 I solved it with setenv "HOME" "/tmp"

How does one configure sudo?

>>27182 I never did it, but the only thing I found in the manual is: http://guix.gnu.org/manual/en/guix.html#index-sudoers-file >configuring sudo sounds like >stepsister broke GRUB again It's beyond my comprehension.

It's a large thread and ctrl+f yields little, but sorry if it was already asked. How does one go about installing programs not packaged in guile? Or add packages to guile? Being unable to add to the store makes this difficult

>>27550 >How does one go about installing programs not packaged in guile? There are package managers packed in Guix. There's flatpak, rpm, nix and GNU stow, you can try installing one of these. guix install flatpak Manual installation from the source code can be difficult, as everything in /gnu/store is declarative. >Or add packages to guile? First you setup the git repository: http://guix.gnu.org/manual/en/html_node/Building-from-Git.html#Building-from-Git When it's done you can follow the packaging tutorial: http://guix.gnu.org/cookbook/en/html_node/Packaging-Tutorial.html#Packaging-Tutorial You can see many examples in the source code in the ./gnu/packages/ directory. But it isn't always easy - some packages are fuarrrked up by developers and provide nonstandard build system configurations, directories, etc. >Being unable to add to the store makes this difficult You should never modify the store manually. What provides reproducibility and security is the store being managed by Guix and only by it.

Guix is really great technology, but I don't think it will ever see wide adoption outside of servers before someone creates a non-GNU distro based on it. Free software autism zealotism results in a very user-hostile documentation, they have to recommend software that will simply not work for 90% of users. Few people want to use a distro where they have to essentially maintain their separate fork anyway in order to use the software they need, like linux, firefox, steam, spotify, matlab (no, octave isn't good enough) etc. But to me it looks great. I really want to try running it on an old thinkpad. Maybe coreboot it as well. But I still don't think it will reach critical mass before a non-GNU distro pops up.

>>27696 >Guix is really great technology, but I don't think it will ever see wide adoption outside of servers before someone creates a non-GNU distro based on it. GNU is not Windows nor OS X. We do not seek wide adoption by selling users' freedom. The claim about Guix being used only by servers is false. Guix is mainly used by GNU freaks as their 100% libre distribution. >Few people want to use a distro where they have to essentially maintain their separate fork anyway in order to use the software they need, like >linux Did you read the april fool's post and didn't notice? Support for Linux-libre stays there http://guix.gnu.org/blog/2020/a-hello-world-virtual-machine-running-the-hurd/ And there are unofficial guix channels for nonfree linux. >firefox, steam, spotify, matlab (no, octave isn't good enough) etc. Mostly proprietary spyware, available through flatpak. No need to maintain a separate fork, Guix provides easy way to add channels: http://guix.gnu.org/manual/en/html_node/Channels.html#Channels But for now there are small chances of getting non-GNU Guix distribution, the reason is Guix's advantages are incompatible with proprietary software. Guix provides transparency, reproducibility and freedom, whereas proprietary software the opposite. GNU/Linux distributions seeking wide adoption at the cost of user freedom don't care about the advantages of Guix which are minor compared to the convenience brought by binary packages and proprietary software. That's why distributions such as Mint, Manjaro, Elementary, etc. invest their resources in software like Wine, Steam, Spotify or Snap packages.

>>27710 > non-GNU Guix distribution NixOS. > the reason is Guix's advantages are incompatible with proprietary software. https://github.com/NixOS/patchelf > Wine, Steam, Spotify or Snap https://nixos.org/nixos/packages.html?channel=nixos-20.03&query=steam > channels https://gist.github.com/edolstra/40da6e3a4d4ee8fd019395365e0772e7

>>27711 What's your point? Nix isn't a Guix distribution. Guix was created as an alternative to Nix with more emphasis on Free Software (and Scheme).

>>27711 >NixOS. NixOS is cool and all, but it uses systemd, uses a poorly designed domain-specific language, promotes proprietary software and as the above poster said, it isn't a Guix distribution. >https://github.com/NixOS/patchelf Patching proprietary software is illegal and my point was more about proprietary software being insecure, malicious and not reproducible by design.

>>27715 Yes, it is not a Guix distribution, however it shares the same functional ideas through and through. And it just doesn't feel right not mentioning it all, while blatantly lying that... > Guix's advantages are incompatible with proprietary software ...when they are, and if patchelf is illegal then chroot is not. Also... > systemd > a poorly designed domain-specific language ..always crop up (I actually was waiting for these two) when somebody brings up NixOS, while being silent on all advantages it has over Guix. And don't get me wrong I like Guix, as well as I like Haskell, Nix, Elm, PureScript, ML and other functional kawaii. But such attitude is just disgusting. And let's be real. > systemd Is convenient, especially for desktop users. Most of the time you simply do not care that systemd even exists, and if you do, the reason most likely is, you want to read journalctl or restart some stray services, which is easy. > a poorly designed domain-specific language In which way? It looks exactly like JSON, and if all you want is to activate a bunch of services and add some packages, you don't need to know it at all. In this way it is quite elegant. Yes, lots of things break apart when you want to do something non-trivial, but even then, you usually do not use Nix as much as Bash with ${} inside. Usually you can just copy-paste some script, tweak it a little bit and it will work. If you are talking about this... https://gist.github.com/edolstra/29ce9d8ea399b703a7023073b0dbc00d ...then sure, but it is acknowledged and already addressed. As well as channels are. > promotes proprietary software And this is just ugly. It does not, and you know it, yet you blatantly lie about it. I wonder why? First you need to activate unfree packages in a config file, as well as you might broken or unsecure packages. And second, you are free to publish and install whichever packages you want, without any hindrances, which is a freedom as I see it.

>>27716 It's difficult to argue with fsf zealots, because of their deep emotional investment in the free software ideals. I'm guessing simply mentioning proprietary software in the documentation is considered a promotion. The open source community tends to be more pragmatic. I'd like to be free from proprietary software, but sometimes that just isn't feasible, either due to a job/class requiring use of a proprietary software package, or due to hardware requiring firmware blobs to function. I can either live like rms, or take a pragmatic approach. IMO, zealotry always cripples the mind, no matter what you believe. It hinders you from thinking in flexible ways, and turns you into a frustrating brick wall in an argument. That isn't to say GNU is never pragmatic, they do maintain a Windows port of Emacs for the people who are forced into that platform.

>>27716 >Yes, it is not a Guix distribution, however it shares the same functional ideas through and through. Well, you didn't make it clear and caused a misunderstanding. You responded to "non-GNU Guix distribution" with "NixOS". >And it just doesn't feel right not mentioning it all, while blatantly lying that... Could you please not use such emotional language to refer to my post? Having different opinion than someone isn't lying. >...when they are How is proprietary software reproducible, secure, user-controlled or trusted? It is possible to patch proprietary software to use dependencies built in declarative and reproducible manner, but that's all. The binary was still built with an unknown compiler, uses proprietary dependencies and is proprietary itself. Modifying proprietary binaries is also often illegal and it's really hard to check what the software actually does. You can't make it reproducible if you don't have the source code. >while being silent on all advantages it has over Guix. Name them. I only know disadvantages, because I've never used NixOS. >systemd >Is convenient, especially for desktop users. Maybe it is, but I used none of its features on a daily basis before switching to a non-systemd distro. All I do is "sudo herd stop/start <service>" and I see no reason to have the entire platform and feature creep in my PID-1. Having RedHat's dick in my distro is not nice. >In which way? The problem is, instead of using a ready, well documented and mature programming language, they decided to reinvent the wheel. From the list you posted I can see they're wasting time on making their syntax and language features work. That's the problem - it's another language you have to learn in order to use their distribution, make packages, but they're the only things you can use Nix language for. With Guix you have many Scheme libraries and standards already available and the syntax is extensible by design through macros for example. >It does not, and you know it, yet you blatantly lie about it. Emotional language. It contributes nothing to the discussion. >First you need to activate unfree packages in a config file, as well as you might broken or unsecure packages. By providing nonfree packages easy to install you treat it like there's nothing wrong with the software. Is there a warning somewhere in Nix documentation? Something like: "This software can violate your privacy, restricts modification and distribution"? If not, then Nix community by providing support for malware in their distribution, promotes this software. They provide support for it and make it easy to install and by this they help it grow. In my opinion this equals to promotion. It's like giving kids instructions where they can find drugs and not warning them drugs are bad for health. >And second, you are free to publish and install whichever packages you want, without any hindrances, which is a freedom as I see it. You're right, this is freedom, but just because you're free to do something, it doesn't mean it's a right thing to do. You can install it and you have the freedom to do this, but be prepared the software will exploit you and someone will restrict your freedom. Guix people chose the path of making a 100% free distribution, because they believe the world should be free from software imposing restriction on users. Guix community doesn't provide support for software restricting freedom. You're free to install it yourself, but be prepared to take all the inconvenience of making it without anyone's help.

>>27717 If you would have seen the zealots in the soviet union in the 60's and 70's.. the paralelles are disturbing to say the least.

>>27719 > Having different opinion than someone isn't lying. If it disinforms people, than it might as well be, opinion or not. > reproducible Let's first define what reproducibility is - an ability to install and use a package in different environments on different machines. Next, can you wrap the binary which was... > built with an unknown compiler, uses proprietary dependencies and is proprietary itself. ...with all required libraries and make it reproducible? Yes you can. So what should I make of it then? > Guix's advantages are incompatible with proprietary software Nix is just a tool, which does not care what you give it - you can build C code with it, you can store reproducible pictures from internet in your /store/, and you can use it to reliably manage proprietary binaries. There are things with which Guix's advantages are incompatible, say, managing your secrets. And that's still an open question in the community - you don't want your passwords and keys to rollback with your system, and you also don't want them to be globally readable. But, again, it has nothing to do with proprietary software. > You can't make it reproducible if you don't have the source code. So, no, you can. And other points are unrelated to the discussion at hand. > Name them. I only know disadvantages, because I've never used NixOS. Why, quite a few. Number of packages and maintainers. https://repology.org/ Nix-shell and lorri for managing environment of your projects. It is way more easy to install. Integration with cabal and stack for my Haskell projects. Again, I'm not saying that you should use NixOS because of this, and I'm not saying that you can't have all those things on Guix given enough man-hours. But for me convenience is way more important then some abstract ideas of being systemd-free and having a better language. I don't want to discuss systemd, because I already had enough of that - in the end it has nice support, is easy to use for daily routine and helps maintainers doing their maintainer's stuff. > it's another language you have to learn in order to use their distribution It goes both ways, I never really programmed in any lisp-dialect, it is quite foreign to me. And I won't use it for anything other than making packages anyway. > With Guix you have many Scheme libraries and standards already available and the syntax is extensible by design through macros for example. And with Nix you have a JSON-like syntax, which is way more easy to read (at least for me), laziness and purity. Everything else you can get with Bash, which everyone knows and which everyone uses for many other things then for only making packages. But then again, I also use Nix for setting up my environment in Haskell projects, so there are more things to the language too. Just some random configuration - everyone can read it without knowing anything about Nix, which is enough in 90% of cases. And which I consider to be a good design overall. https://github.com/jluttine/nixos-configuration/blob/master/configuration.nix > Is there a warning somewhere in Nix documentation? There is this, which is enough. error: Package ‘vscode-1.44.1’ in /nix/store/9p3s4r06ihysx2l2gi6n18m7kylgy2pq-nixos-20.03.1754.b0e3df2f843/nixos/pkgs/applications/editors/vscode/vscode.nix:40 has an unfree license (‘unfree’), refusing to evaluate. > If not, then Nix community by providing support for malware in their distribution, promotes this software But it is all community-driven, people maintain unfree software for themselves, and if they don't, it will be simply thrown away at some point in the future. My freedom would have been more restricted if I weren't able to read visual novels, because someone decided that Steam is bad for me. > It's like giving kids instructions where they can find drugs and not warning them drugs are bad for health. > but just because you're free to do something, it doesn't mean it's a right thing to do. But we are not kids and can make proper decisions based on our own best judgement. I don't want somebody telling me what I can't do and always reminding me why. It is just propaganda, and I had enough of that. > making a 100% free distribution By restricting others. > You're free to install it yourself, but be prepared to take all the inconvenience of making it without anyone's help. Yes, I also had this experience before - you are free to do whatever you want, but you will be condemned if you do so. I'm more of an opinion that people should be able to eat their poison if they want. Informing them about hazards of using proprietary software is nice (like nix does), however constantly drilling it into people's heads is just propaganda, and this can (and was!) be easily abused.

>>27724 >Let's first define what reproducibility is - an ability to install and use a package in different environments on different machines. This isn't reproducibility, this is portability. https://en.wikipedia.org/wiki/Reproducibility >Reproducibility is the closeness of the agreement between the results of measurements of the same measurand carried out with the same methodology described in the corresponding scientific evidence In other words, making sure the output of processing data will be always the same if the input data is the same. That's why you should avoid telling people they're lying. Lying is telling people something false as true, despite knowing it is actually false. >If it disinforms people, than it might as well be, opinion or not. If I were you, I could tell you you're lying now (which isn't right). >...with all required libraries and make it reproducible? Yes you can. Can you grab all libraries needed to compile Steam or just name it? Good luck with that. Vast majority of the libraries are proprietary software. You can't have reproducibility - the same output, if you don't have the input data and if you can't even repeat the process of compilation. The binaries will differ and act differently, just because they were built using different compilers, in a different environment using different libraries. What you're trying to say here is you can reproduce McDonald's hamburgers just by removing pickles from it. The truth is the recipe is proprietary and ingredients are proprietary and you cannot make the same hamburger. >There are things with which Guix's advantages are incompatible, say, managing your secrets. Guix is a package manager, to managing secrets you use password managers, full disk encryption and brain. >Why, quite a few. >... I agree with this, Guix still lacks some features, because it is just 1.1.0 release. >But for me convenience is way more important then some abstract ideas of being systemd-free and having a better language. Then it's okay. Guix was created for different people with different values. Use what's convenient for you. >Everything else you can get with Bash, which everyone knows and which everyone uses for many other things then for only making packages. If they have to use Bash where Nix language fails to provide features, this means Nix language isn't a good programming language. It is a language good for just few things. Even if it's easy, it could be better. They could use Lua or something designed for the purpose, it would be much better https://www.lua.org/start.html But if people are okay with using yet another domain-specific language and Bash, then I can't say much. Just imagine every program would define their own configuration language (nightmare). >It goes both ways, I never really programmed in any lisp-dialect, it is quite foreign to me. Guix was made for people liking, wanting to use Scheme. It's just not for you, if you don't like Scheme. >And I won't use it for anything other than making packages anyway. The difference is you can create a web server, package manager or process supervisor in Scheme, while in Nix lang you can't. If you use your time to learn a language, why should you be limited to only few applications? >There is this, which is enough. Good they have this at least, but for new users this can be not enough. There was the time I didn't know about software licenses and accidentally installed nonfree packages, thinking they're fine and ethical, just because they were included in the repository of a popular distribution without any notice. The EULA of a game was literally obligating you to give up all rights to all content you created in the game. >But it is all community-driven, people maintain unfree software for themselves, and if they don't, it will be simply thrown away at some point in the future. Okay, then that's fine. >My freedom would have been more restricted if I weren't able to read visual novels, because someone decided that Steam is bad for me. You don't seem to know how copyright works. You're always free to read copyrighted work, unless it's software. And no one forbids you to install Steam on Guix, there's no DRM or something, they just don't support you in your decisions. On the other hand Steam restricts what you can do with software you use. > making a 100% free distribution > By restricting others. Haha, good one. Not packaging someones malware is not restricting freedom of users of that malware. It's just not providing support. >however constantly drilling it into people's heads is just propaganda, and this can (and was!) be easily abused. You're right. This is a known mistake in the free software community. It is easy for people thinking they're right and they're doing something good to start being aggressive. That's why I try not to use emotional language while talking about it, guess it fails sometimes. But Guix community is nothing like usual freetards, I didn't encounter anything negative personally.

>>27726 > This isn't reproducibility, this is portability. Oh, no, by environment I meant not your system, but your packages, configs and other mutable things on your system. Here is a definition from Nix webpage mine was loosely based on. > Reproducible > Nix builds packages in isolation from each other. This ensures that they are reproducible and don’t have undeclared dependencies, so if a package works on one machine, it will also work on another. And if you have the same binary, and you wrapped it with the same packages, you would get the same result, no matter what you have on the system. I do agree that you might not be able to start a proprietary application after some time, say they might force you to use newer versions and you won't be able to do anything about it. But that behaviour would be intrinsic to an application, not to the libraries it uses. > Can you grab all libraries needed to compile Steam or just name it? I actually can show all libraries Nix knows about. > Guix is a package manager Oh, but they are both way more than simple package managers, With them you can declaratively configure your whole system. You can enable encryption of your disk, or you can even declare your passwords in the main config. However it is clearly a bad idea. And there is now (or maybe ever?) no meaningful way to do it properly. The other problem is how to store suid executables. You can't put them into store, because the user who does it has no privileges whatsoever. > Just imagine every program would define their own configuration language (nightmare). Actually we already live in a menagerie of programming languages and in a zoo of config files. But I see why one might want to use a single tool for many things, I would also want to use Haskell for pretty much anything. And it has nice support for DSLs too! > Lua But it is not functional, it is not lazy, it is not pure, it would be a total mess, really. Nix Language works as a rigid framework to which you can add anything you want, whether it is bash, make or something else - nix does not care. It is just that bash is already used in *nix community as a universal glue to compose your whole system. And with nix it works way better. There is a nice separation of concerns. And everyone (well, almost) is happy. > It's just not providing support. Even if users want it? Well, I think it is just a matter of perspective. And just to be completely clear, there are so many bad things which happened in software industry (UNIX is arguably one of those) - Guix, on the other hand, is completely different. I don't think there should be just one right distribution to rule them all. It just feels disingenuous when one says that systemd and Nix language are inferior to what Guix offers, things are simply done differently. Which is nice! %%An image with dependencies was too big and I couldn't upload it, but believe me it is there.%%

>>27727 Sorry to interrupt your autistic discussion in full flow, but what is the nix language like to work with? How similar is to haskell? Why didn't they just use that for the config? I'm planning to try one of guix or nixos next week, and having trouble deciding between them. I'm more familiar with haskell than lisp, but on the other hand I live almost entirely in emacs, which benefited massively from having a full language as the config that resulted in some neat extensions.

>>27737 It highly depends on what you want to do with it. If all you need is to flip toggles in your config than Nix is fine. If you want to maintain a non-trivial package, than it might be really painful - most of the documentation is scattered throughout different man pages and answers on the official discourse, and you just usually end up looking how other peoples do stuff and copying it. It also brings lots of boilerplate for not-exactly-package-related-things. Just look at this monstrosity, and I have something very similar set up. https://github.com/Gabriel439/haskell-nix/blob/master/project4/release3.nix However if all you need is configure, make, make install, than it is pleasant, just look at some random package. https://github.com/NixOS/nixpkgs/blob/master/pkgs/games/atanks/default.nix There are lots of functions for other builders, languages, etc. But again, you usually learn about them by looking at examples. It is similar to Haskell because it abuses laziness through and through - it ties the knot of all packages and configs on your system, it uses open objects to allow late rebinding and you close it with fix function later (not directly). It has recursive let bindings too, also everything is immutable by default. So if you are familiar with these concepts then it should be simple. > Why didn't they just use that for the config? It uses open sets and reflection for almost everything, while you can do something similar in Haskell, it would be super-super painful. And it would also bring GHC as a dependency, which is quite big and scary. > guix or nixos Nix is easy and convenient. Some people might not agree, but for me it is. It has wider adoption, which results in more packages, which results in more convenience again. And it is simple to install - you just need to partition your disk and call nixos-install, or something. I also use lorri a lot now. Don't know whether Guix has an alternative. https://github.com/target/lorri

>>27745 Guix you can easily install non-free software through, granted it will be source and not a binary but things like 'proprietary' linux kernal or whatever else is easy. >>27737 Look up guix papers on google scholar, it will give you a crash course, then read the Nix paper they're essentially the same thing except guix uses guile scheme. I actually prefer the Nix syntax but on the other hand it's easier to parse guile sexp so you can then programatically deploy a ton of drone servers from emacs.

>logo is a vagina

>>27727 I didn't write you back because I was too tired, sorry. Not that you wanted me to lead this autistic discussion any further :P >Oh, no, by environment I meant not your system, but your packages, configs and other mutable things on your system. I don't really remember what was my point here, so I'll just skip this. >And if you have the same binary, and you wrapped it with the same packages, you would get the same result, no matter what you have on the system. If you treated a proprietary binary as your source, then given the same binary, the processing using Guix/Nix will be reproducible. Let me reword myself - what I mean by proprietary software being incompatible with Guix is not its technical abilities, but philosophy. Guix focuses on reproducibility, bootstrapping and hackability (by using Scheme) to liberate its users and give them security. By adding support for proprietary software, the mission of the project gets compromised, as often proprietary software is insecure and restricts its users. That's what I mean by the incompatibility. I don't really remember now if this was my point in the beginning of the discussion, or my mind changed over time... >I actually can show all libraries Nix knows about. Okay, maybe Steam was a wrong example, because AFAIK it's based on chrome/chromium and many components are indeed freely licensed. Still there are some proprietary, non-reproducible parts, which can act differently between releases - that's the minus of proprietary software you can't overcome. You also can't control how a proprietary package is packaged - it is known in Guix community, tarballs generated from Github are not reproducible - they get regenerated sometimes. That's why you can't be sure if downloading the same release of a package from a proprietary distributor twice will give you the same result. This breaks reproducibility and undermines trust. >Oh, but they are both way more than simple package managers, With them you can declaratively configure your whole system. >You can enable encryption of your disk, or you can even declare your passwords in the main config. However it is clearly a bad idea. And there is now (or maybe ever?) no meaningful way to do it properly. That's not exactly a part the package manager - there's Guix and Guix System. There are two separate directories for package manager - the ./guix directory contains the code of the package manager and the ./gnu directory contains the code of the distribution, package and service definitions, installer, bootloader configurations, etc. If there was a need to add a feature not related to package management, a proper Scheme code would be written to handle it. The whole system configuration of Guix System is actually a Scheme code with handy syntax - from there already written programs and services can be executed. So even if Guix is bad at "keeping your secrets", because it is only a package manager, the whole system configuration could be extended to be good for the task. http://guix.gnu.org/manual/en/html_node/System-Configuration.html This is all in one repo and under one "Guix" name, but these are actually different things. >But it is not functional, They mention functional programming on their website: https://www.lua.org/about.html >it is not lazy, it is not pure, it would be a total mess, really. Didn't use it, so you may be right. I mentioned it, because it's advertised as "ideal for configuration". >Nix Language works as a rigid framework to which you can add anything you want Could you make a procedure generating custom packages? In Guile Scheme it is easily doable. I wonder how flexible is Nixlang. Would it be hard to write a bash script generating custom packages? >It just feels disingenuous when one says that systemd and Nix language are inferior to what Guix offers, things are simply done differently. Which is nice! It depends on what you want - some users don't like systemd and want 100% free distribution, others want to use some proprietary software and systemd because it just works (even though it's bloated). That's why I mentioned it. For some users systemd would be a too big disadvantage to go for Nix. If they wanted to try a functional distribution, but still wanted to use some proprietary software, Guix with a nonfree channel could be better for them than Nix. Hence my reaction to you responding with "Nix" when the other lainon said "Guix distribution". The lain that triggered my response said: >Few people want to use a distro where they have to essentially maintain their separate fork anyway in order to use the software they need, like linux, firefox, steam, spotify, matlab (no, octave isn't good enough) etc. This whole discussion started, because Guix is made by people who want it as it is now for people who want the same and I didn't see a reason to change it. Maybe if Guix was a game store just like Steam or whatever, the adoption would be wider, but that's not the point of the project. My point was also that there's no need for a separate Guix distribution, because adding support for proprietary software doesn't require forking - just an unofficial channel would do the thing. I hope this makes sense.

I'm coming back to Guix now that I can run it on my T420s. Wasn't possible before with the original Intel wireless card.

>>29171 you replaced the wireless card? or started using the nonfree driver?

>>29178 I was able to replace it after corebooting it (effectively getting rid of the BIOS whitelist)

>>29186 How did you Coreboot the T420s? I have one and would like to do this as well.

Why does this emblem look like the emblem of Villars chocolate?

>>29193 I don't know, let's stay with a vagina >>28218

>>17153 >Stockholm Syndrome with people coming from Debian/Gentoo/Arch I don't think you know what that is. Debian and Gentoo are great, specifically Debian. I have yet to find something that's as stable, easy to install and use while still being very minimal, as Debian. Plus it stays true to free software by default.

>>29203 >Debian and Gentoo are great, specifically Debian. Not the lainon you're replying to, but I find Debian complicated to install and it gives me poor desktop experience - outdated (stable) software. Gentoo seems to be better, thanks to its source-based model and rolling release, but I can't forgive it not having an installer - I failed to install it manually on my EFI machine, with fully-encrypted SSD drive and dualboot with W10. Guix System is easier to install than Gentoo, even without an installer - the whole system configuration in Scheme is just better than having many files in different directories. Don't know what did they mean by "Stockholm Syndrome" though.

Guix now has authenticated updates. This makes it secure against many types of attacks. https://guix.gnu.org/blog/2020/securing-updates/

>>29207 >but I find Debian complicated to install you have multiple ways to install it, lainon. 1. use the live cd graphical interface (easy and intuitive) 2. use the advanced installer (i don't really remember its name, but it has a blue background) 3. use the graphical installer (it's option 2. with colors and physical buttons and it's a bit easier) >it gives me poor desktop experience you might be right, but it's not debian's fault, it's because of the DE's default theme, but you have the freedom to polish the experience yourself or change the DE with (tasksel) >outdated (stable) software. they have reasons for that (don't suffer from shiny new stuff syndrome): 1. you barely get new features with every update 2. those features might introduce new bugs and security issues and if you don't care about feature but you do care about security issues and critical bugs, well, debian pushes updates instantly for stable, if a critical bug was patched in any software. but if you still want an up-to-date package(s), again you have multiple options: 1. use backports (which have newer versions of many packages, including the kernel). 2. use proposed-updates with backports 3. use testing branch which is very stable for daily use

>>29195 The logo represents the origin of freedom and opportunity, so a vagina makes perfect sense.

Hurd bros and sisters, the userland-driver interface is now considered "acceptable" and will be pushed to master. One step towards reunited GNU system! https://lists.gnu.org/archive/html/bug-hurd/2020-07/msg00038.html Hello, So thanks to Damien's polishing work, we got an acceptable userland-driver interface which I could commit mainstream. I have just uploaded in Debian a GNU Mach version 1.8+git20200710-1 that includes it. We now have to fix netdde and rump into using that proper interface. That's yet another step towards user-land drivers: we already have network drivers, Damien is working on disk drivers. Once we have that working, we will be able to get rid of the old Linux piece included in the gnumach source, that is currently only used for the disk support. Samuel The latest Guix GNU/Hurd images are also available to download here: http://guix.gnu.org/download/latest/

>>29505 Someday x86_64 (or far better, OpenPOWER) will be supported.

Anyone here managed to run tor browser on Guix? I know there's no official package, but maybe someone hacked together the binary from Tor's website to make it work.

There's an article explaining how to build a cluster with Guix. https://guix.gnu.org/blog/2020/running-a-ganeti-cluster-on-guix/ Just imagine connecting your all thinkpads into one, powerful cluster! Maybe I should give it a try.

>>29207 >I can't forgive it not having an installer

>>29296 Yes, not having vulns is good.

>>29932 Last time I checked, they had "you're the installer!" on their website.

why should i install this over freebsd

>>30084 Whole system configuration in one language, reproducible builds, transactional package management, declarative environments? License that saves you from being sued by a patent troll, if you contribute to the project. http://guix.gnu.org/manual/en/html_node/Features.html#Features http://guix.gnu.org/manual/en/html_node/Using-the-Configuration-System.html

>>21180 Nah, I'm not touching this s⁣hit after this. What a bunch of c⁣ucks, lol. RMS rulez!

OMG I CAN'T WAIT

>>30776 Call me skeptical, but what is the point of using a microkernel that is, after more than 25 years, still unfinished?

>>30798 >Call me skeptical, but what is the point of using a microkernel that is, after more than 25 years, still unfinished? Good architecture doesn't get old, good ideas neither. Despite the long time, it still has advantages over kernels used up to this day: https://www.gnu.org/software/hurd/faq/still_useful.html https://www.gnu.org/software/hurd/advantages.html https://www.gnu.org/software/hurd/doc/hurd_1.html#SEC3 >it's compatible >The Hurd provides a familiar programming and user environment. For all intents and purposes, the Hurd is a modern Unix-like kernel. The Hurd uses the GNU C Library, whose development closely tracks standards such as ANSI/ISO, BSD, POSIX, Single Unix, SVID, and X/Open. >it's built to survive >Unlike other popular kernel software, the Hurd has an object-oriented structure that allows it to evolve without compromising its design. This structure will help the Hurd undergo major redesign and modifications without having to be entirely rewritten. >it's scalable >The Hurd implementation is aggressively multithreaded so that it runs efficiently on both single processors and symmetric multiprocessors. The Hurd interfaces are designed to allow transparent network clusters (collectives), although this feature has not yet been implemented. >it's stable >It is possible to develop and test new Hurd kernel components without rebooting the machine (not even accidentally). Running your own kernel components doesn't interfere with other users, and so no special system privileges are required. The mechanism for kernel extensions is secure by design: it is impossible to impose your changes upon other users unless they authorize them or you are the system administrator. It is just a matter of promoting it more before it starts working.

>>30810 > Unlike other popular kernel software, the Hurd has an object-oriented structure that allows it to evolve without compromising its design. This structure will help the Hurd undergo major redesign and modifications without having to be entirely rewritten. Well, Windows is object-oriented too, and it is a complete failure, as OO in general.

>>30985 >Well, Windows is object-oriented too, and it is a complete failure, as OO in general. NT's failure is mostly because it is proprietary which automatically makes it incompatible, built to rot, unstable and unscalable for users by design. >is a complete failure, as OO in general. OOP failed, because people are thinking about it in terms of C++ or Java, but the original idea of OOP was about passing messages between objects: https://en.wikiquote.org/wiki/Alan_Kay >Actually I made up the term "object-oriented", and I can tell you I did not have C++ in mind. I believe OOP done correctly is good.

can i have CDE on guix? I'll install it right away if so

do i need to know lisp in order to use this distro?

>>31055 No, but you’ll want to learn it if you want to do anything with it. You’ll pick it up, it’s not hard - you can do a lot of expressive stuff with config on Guix because it’s built in Lisp, but you don’t have to do any of that if you don’t want to, the configuration language is a DSL in itself which is lispy in syntax but doesn’t require a thorough understanding of Lisp.

>>31054 Quick search gives no good results. CDE is a DE, right? >>31055 >do i need to know lisp in order to use this distro? Only basic syntax to do configurations, the package and configuration interface is highly abstracted. There's the Scheme crash course in Guix's Cookbook: https://guix.gnu.org/cookbook/en/html_node/Scheme-tutorials.html#Scheme-tutorials Knowing Scheme makes it more fun though.

>>31055 You should, but dob't have to know Scheme, or more specifically Guile. In the same way as you show know they shortcuts in the software you use -- it's in your interest, and it's not that difficult. >>31054 Don't see why not, even if there's no package you can build it yourself (or modify an existing recipie to build it via guix).

I was waiting for Guix 2.0, but 1.1 might be what the guy meant who told me to wait. Any huge upgrades coming which would need me to do a reinstall? I just want a distro with all features that works and can be upgraded, no more distro hopping. How much hazzle is it for a laptop, when I have some experience with a version of Lisp and Linux?

>>31228 >I was waiting for Guix 2.0, but 1.1 might be what the guy meant who told me to wait. 2.0 will probably be the release where the Hurd is usable enough, 1.1 was already released some time ago, what you could be waiting for is 1.2 or 1.1.1 depending on developers' vision. >Any huge upgrades coming which would need me to do a reinstall? Guix System is rolling-release. The release numbers are only informative and if you're running Guix already all you should do is "guix pull && guix upgrade && sudo guix system reconfigure /etc/config.scm". Or do you mean switching distros? As mentioned below the support for GNOME and general improvement of some packages is coming. >I just want a distro with all features that works and can be upgraded, no more distro hopping. There are still some features missing, like a good support for desktop environments, but GNOME is in progress. Personally Guix System stopped me from distro hopping, but at cost - Linux-libre has weak hardware support and your GPU is going to work without 3D acceleration. What made me stop distro hopping is I don't have to care about major version upgrades, I just reconfigure the system and enjoy - no need to reinstall system and set up million different things I never remember. >How much hazzle is it for a laptop, when I have some experience with a version of Lisp and Linux? If the laptop is low powered software upgrades can be a problem - Guix often compiles the packages from source code, if a binary substitute is not available yet. Experience with Lisp helps much, but general experience with "Linux" doesn't - you don't do things like creating a user or a group manually, you don't edit fstab, etc. For configurations there's the whole system configuration interface https://guix.gnu.org/manual/en/html_node/System-Configuration.html So unless your laptop is powerful and you're ready to install nonfree Linux, you should rather use Guix as your package manager on other distribution, not on Guix System.

>>31245 Thanks, I also have enough to reinstall everything without knowing what it was. Though, some of it could be solved by improving the distros, like keeping a list of all programs I had. But I also don't want to deal with C++ or C, I prefer Lisp/Scheme. Yes it's meant as a desktop, so this should be working. On that note, how difficult is it to get around the restrictions by the devs? I mean, I can install packages from other distros, or not?

>>31247 >Though, some of it could be solved by improving the distros, like keeping a list of all programs I had. On Guix there's the /etc/config.scm file, where you can put the list of packages to be installed: (packages (append (map specification->package '("nss-certs" "icecat" "font-dejavu" "libreoffice" "mpv" "gimp" "emacs" "emacs-paredit" "emacs-auto-complete" "gnupg" "htop")) %base-packages)) >But I also don't want to deal with C++ or C, I prefer Lisp/Scheme. Same. >On that note, how difficult is it to get around the restrictions by the devs? There are no any restrictions, it is just a philosophy the project follows - developers do not provide software that denies users' their basic freedoms. There's no community-driven effort to provide unethical software, but you're free to install it yourself. >I mean, I can install packages from other distros, or not? Guix provides channels - an easy way to install unofficial software, there are some channels with nonfree software also, just search for "guix nonfree" on github and other places. https://guix.gnu.org/manual/en/html_node/Channels.html#Channels You can also install other package managers using Guix, there's Nix and Flatpak in the repository, which already provide some nonfree software. One of Guix contributors has instructions on how to set up nonfree Linux, just scroll down: https://ambrevar.xyz/guix-advance/index.html

>>31248 Oh wow, this sounds great. Have no time currently, however. Just got my 3D printer. But soon. Maybe the Gnome or any desktop support is there till then.

>>31245 Any idea how to upgrade only those packages that have binary substitutes available?

>>31265 >Any idea how to upgrade only those packages that have binary substitutes available? Currently the only solution is to just wait some time before major changes. The build farm is powerful enough to build derivatives quickly. Practically this means you run guix pull one day and guix upgrade; sudo guix system reconfigure /etc/config.scm two days later. There's also the do-not-upgrade flag $ guix package --upgrade . --do-not-upgrade ungoogled-chromium https://guix.gnu.org/manual/en/html_node/Invoking-guix-package.html#Invoking-guix-package You can use it to skip upgrading bloatware. Generally avoid upgrading packages like web browsers - icecat, webkit, chromium and libreoffice. Upgrading linux-libre also takes some time and resources, but if you have at least 3GB of RAM, it should do fine. If you have little RAM memory, you can also consider turning off some services like X. The way of doing this is to switch to a tty, for example by using Ctrl+Alt+F3, login into a session and run sudo herd stop xorg-server This stops your GUI and gives you additional memory and computing power to build software. Remember to add the cores flag when pulling, upgrading and reconfiguring, say you have 2 cores then you type guix pull --cores=2; guix upgrade --cores=2 etc. >>31263 >Maybe the Gnome or any desktop support is there till then. Part of the GNOME support was just merged to master branch, it can be ready soon.

>>31081 yes, common desktop environment

>>31282 I could try making a package definition for you if you asked nicely. But I'm a lazy fuarrrk, so good luck with that.

>>29505 Holy soykaf, I was waiting forever for images to be released and it happens and I miss it.

>>31245 use the nonguix channel

>>31248 >There's no community-driven effort to provide unethical software, but you're free to install it yourself. Yes there is. https://github.com/ajgrf/nonguix

Ever wanted to run a Hurd cluster? https://github.com/BrentBaccala/hurd https://gnu.hurd.bug.narkive.com/RX8i0fJZ/hurd-lecture

Is KDE plasma available yet?

>>31313 >Is KDE plasma available yet? No >31313 Cool numbers

I have used Guix system for a while and i liked it, but I'm going to wait until i get better at programming/scheme. In the meantime I'm going to use Debian and put guix on to it, having a stable distro with a rolling release package manager seems pretty sweet.

I want to use this so badly, but I don't have any compatible hardware.

>>18649 >NixOS uses systemd, which is really bloated and hard to understand, wheres GuixSD uses the GNU Shepherd. Perhaps it's quite late to reply to this, but whatever. If you are concerned about systemd's complexity, you probably shouldn't use either Nix nor Guix.. Just have look at their code base. At least Nix removed the creepy Perl dependencies recently, but nevertheless.

>>27182 >>27185 (define %sudoers-specification (plain-file "sudoers" "\ root ALL=(ALL) ALL %wheel ALL=(ALL) ALL)) ... (operating-system ... (sudoers-file %sudoers-specification) ... >>27722 What a lazy and vague comparison.

Has anybody been able to setup EXWM on guix? I know it has been packaged but when I try to login through gdm it just goes back to the login page

>>31815 Did you add the proper service type to your config.scm? Just installing it won't work.

Anyone know anything the gash shell being worked on? Is it meant to replace bash or what?

>>31888 >Anyone know anything the gash shell being worked on? Is it meant to replace bash or what? The gash shell is mainly used for bootstrapping in GNU Mes. From the manual: >Gash is a POSIX-compatible shell written in Guile Scheme. It has two >goals: bootstrapping GNU Bash and exposing the shell to Scheme. I tested it and it does basic stuff, but isn't convenient. Somewhere either on their website or in the manual I found the info Gash is currently slow and lacks some features, but they plan to make it usable. I didn't see much activity recently though: http://git.savannah.nongnu.org/cgit/gash.git

My Guix installation stopped working due to some GPU firmware magic. The pathetic state of modern proprietary hardware forced me to do this. Fortunately there's nonguix channel for proprietary software: https://gitlab.com/nonguix/nonguix The packages are really fresh and the channel is active - they even recently added a package for Steam and it runs in a container, meaning you can play proprietary games without compromising your system (that much). Finally I can play Minetest with 3D acceleration and fonts are rendered properly. I hope there will be some libre hardware available soon and I don't know why didn't GNU nor the FSF considered making a fund for development of libre hardware. They made the FSDG and tell distribution maintainers not to add any nonfree software, not even firmware, but they have no alternative, which is sad.

>>31888 if you want a lispy shell there is always emacs' eshell.

>>31819 Idk it was weird, removed and reinstalled it and now it works fine

>>32004 There are already some projects underway to get fully libre graphics and such going, but they are still stuck in development.

>>32004 Sounds neat, but one problem is that the nonguix channel is hosted on GitLab. At least it should be hosted on Codeberg or some other gitea instance.

>>33021 Mirror on Snopyta Git and/or Disroot git.

>>16956 Is this recommended for new Linux users and/or non-programmers? Does using this make the bulk of crucial Linux noob tutorials useless?

>>33026 >Is this recommended for new Linux users and/or non-programmers? No. I wouldn't recommend installing it for someone not willing to learn Scheme before running it at this point: no GUI for the package manager, only basic integration of DEs, you need to know guix commands to work with the system, basic Scheme knowledge is required to add services such as printing (CUPS) or desktop environments. Guix System requires at least some knowledge about GNU/Linux to use it. Documentation is pretty good, but you have to know Scheme first to be comfortable with it. As a new user, you can try installing Guix the package manager on your distribution of choice (it is currently broken on Fedora, due to some SElinux policies). >>33021 >Sounds neat, but one problem is that the nonguix channel is hosted on GitLab. It shouldn't matter from user perspective much, if you're not willing to contribute. Commits are authenticated and fetching from git doesn't require any nonfree software to work - you just connect using standard protocols (git, http(s)).

>>33026 >>33031 Forgot to answer: >Does using this make the bulk of crucial Linux noob tutorials useless? Yes, the whole system configuration mostly abstracts the underlying GNU/Linux logic. For example you shouldn't add users manually or edit fstab. https://guix.gnu.org/manual/en/html_node/Using-the-Configuration-System.html

>>33031 >>33032 Cheers.

>>31649 when did they remove Perl? I’m on the nixos Unstable channel and still have a bunch of Perl bullshit in my store

>>16956 this seems like a good place to share this thing I found a systemd alternative, s6-rc https://github.com/skarnet/s6-rc compared with the others https://skarnet.org/software/s6-rc/why.html its available in the guix repository would like to hear lains thoughts

>>33591 no mention of gnu shepherd..?

>>33591 >>33593 The Shepherd's integration in Guix System is just too good to use anything else.

why Guix/NixOS if you have ansible + ZFS ?

>>33596 ZFS solves a completely different type of problem. Snapshots are just a side-feature. Ansible is only as good as the roles you give it, and those roles do not keep track of the chain of dependencies, or the exact reproducibility of output, nearly as much as Guix does. Most roles just point at the most recent version of something.

>>33591 I haven't had a single problem with runit yet. Been running it for 3 months.

>>33641 he said runit was very good, just not the best >>33591 >>33593 >>33594 https://skarnet.org/software/s6-rc/why.html so, I emailed the creator of this review to see why he hadn't included shepherd and he was kind to answer: ------------------------- ~ska: I didn't include shepherd in my comparison because it's not a generic service manager. It's written in Guile, it's configured via Guile modules, it basically depends on a wholly different (and more complex) ecosystem than traditional service managers; it may integrate well within Guix, but is a lot less portable to other distributions. From a usability perspective, that is too big a limitation for me to even consider it. The fact that it's not written in C also lengthens the code paths a lot and makes the Guile interpreter a SPOF for all the services it manages. Shepherd seems to be presented as the init system for Guix, but it is not an init system: it sends its logs to syslog, which implies that it needs a syslogd service already present when it starts. Even if it is run as pid 1, it needs some preparation beforehand. Shepherd is a mix of process supervisor and service manager, and its authors do not seem to understand the difference between those two things. There does not seem to be a thought-out service model either: shepherd treats a one-shot as a service that is started and stopped when it exits. This is problematic when the user tries to define an init sequence and dependencies between services. A cursory look at https://www.gnu.org/software/shepherd/manual/html_node/Design-decisions.html shows me very naive thoughts (about why to use a daemon vs. not using one) and confused ramblings about dependencies ("we would not want to terminate it. Well, maybe we would...") that do not address the fundamental issues to solve when managing services. Maybe shepherd has evolved since then, but the initial dmd design does not look fully thought-out. All in all, I suppose shepherd has a place in Guix, but I do not consider it either a serious alternative for Linux in general, or a well-designed program overall. Which, unfortunately, is true of most GNU programs. Hope this helps ---------------------------

>>33767 >Shepherd is a mix of process supervisor and service manager, and its >authors do not seem to understand the difference between those two >things. [etc.] Haha, I kinda see what the author means, but I think this attitude is what brought systemd into existence in the first place. Poettering also didn't think much about the design and therefore was successful. Guix and Shepherd are going to outlive all these weird init "replacements" just because they actively develop their own distribution, rather than trying to patch someone's garbage like Devuan does.

Guix improved its Hurd support and now provides some cool tools for development. They added a new "childhurd" service which is a VM with Guix GNU/Hurd, it allows cross-building software for the Hurd from Guix System on GNU/Linux. https://guix.gnu.org/en/blog/2020/childhurds-and-substitutes/

>>18650 Just make a git repo with your patched st, that's the sanest way to manage patches anyway.

I switched to guix about a week ago, and so far the biggest pain point has been rust development. Rustup doesn't work on guix, and they don't have nightly rust packaged. You can't even get a language server running.

>>33971 >Rustup doesn't work on guix, and they don't have nightly rust packaged. I believe having nightly Rust is not necessary unless you're developing for Rust. Guix is a rolling release, source-based distribution. Packaging Rust nightly would trigger rebuilds all the time making the build farm busy, it would also decrease reproducibility.

>>33973 There are a lot of rust projects that use nightly only features, or crates that only compile on nightly, including one that I want to work on. Also, not having RLS makes development somewhat impractical even with stable. It's hard enough keeping the borrow checker happy without having to use a compile cycle for every mistake. I get that there are reasons why they can't package nightly, but rust is pretty popular on here so I think it's worth noting just in case anyone else was hoping to use guix for development.

>>34273 >There are a lot of rust projects that use nightly only features, or crates that only compile on nightly, including one that I want to work on. You can make a rust nightly package for development on your Guix install, I could even try writing the package for you if you asked nicely. But I still think using nightly features for anything is bad design and if Rust developers actually need them, then it only shows how immature Rust is. >Also, not having RLS makes development somewhat impractical even with stable. It shouldn't be hard to package RLS I think. >I get that there are reasons why they can't package nightly, but rust is pretty popular on here so I think it's worth noting just in case anyone else was hoping to use guix for development. Maybe you can ping Guix developers. I'll think about a solution for this problem.

>>34273 >>34279 A simple-enough interim solution would simply be to have a Guix rust-channel so that constant rebuilds aren't a problem on the farms. If at a later point that becomes less problematic, there's no reason it couldn't then be merged in to the main guix-channel.

>>34273 I just tried one thing, but unfortunately it isn't that simple lain@navi ~$ guix build --with-branch=rust=master rust-openssl guix build: error: the source of rust@1.39.0 is not a Git reference Packaging Rust in the reproducible and bootstrappable fashion is really hard, because it uses the train model (a release based on the previous). That's why Rust is rather built from tarballs than from git.

I found a peertube channel with a bunch of guix tutorials, might be of some interest here https://video.hardlimit.com/video-channels/the_gnu_guy/videos?a-state=42

So far, how is Guix (with nonfree repos) compared to NixOS as of the time writing this post?

>>35092 guix is cooler and scriptable

Would it be possible to make something like Emacs that uses Guix as the package manager?

>>35402 >Would it be possible to make something like Emacs that uses Guix as the package manager? Guix has many emacs packages and there's emacs-guix to provide interface for Guix inside Emacs. https://emacs-guix.gitlab.io/website/manual/latest/html_node/index.html#Top

>>35406 I was thinking something like that Haskell IDE that uses Nix.

I wish Gash was a radical non-POSIX compliant shell, that would have been pretty cool.

Does Guix have a proper sandboxing and permission management system like Flatpak?