social.dk-libre.fr is a Fediverse instance that uses the ActivityPub protocol. In other words, users at this host can communicate with people that use software like Mastodon, Pleroma, Friendica, etc. all around the world.
This server runs the snac software and there is no automatic sign-up process.
「 Lix is built by a team of open-source volunteers – and exists to provide an alternative to the commercial interests that have long plagued both upstream CppNix and corporate-authored forks. We’re proud to stand by our open conflict of interest statements, and proud to listen to community voices on issues of sponsorship, direction, and moderation 」
#Guix has already raised half of its annual budget 👇
https://guix.gnu.org/en/blog/2025/update-on-the-guix-fundraising/
This is great but the journey’s not over!
If you can, if you care about #Guix (or #Nix!), about #FreeSoftware and independent projects, consider donating. 🤲
Nixlang is really hard to learn, I wish they had used TOML instead /s
I'm in the process of write my second flake.nix file and so far it goes smoother than what I feared 😅 .. BUT!
I'm packaging a rust app with a gui, so needing X11 and a bunch of libX* , and I'm currently in this loop:
> nix build
[wait 10mn for cargo to build from scratch]
> ./result/bin/my_app
Error: libXbla.so not found
> [add pkgs.xorg.libXbla to buildInputs]
> repeat
Is there a way to speed this up? Can I tell nix to cache some/most of what cargo compiles? Or guess the deps quicker?
#nix
ok now I'm stuck 😥
if I try to run my app it crashes instantly with:
> Library libxkbcommon-x11.so could not be loaded.
I've added libxkbcommon to buildInputs.
If I run $ find /nix/store/ | grep libxkbcommon-x11 , it find several instances of libxkbcommon-x11.so , for example:
/nix/store/klv07fh03501djj4q146z6pl7ywc550w-libxkbcommon-1.11.0/lib/libxkbcommon-x11.so
So I guess I have to fix the LD_LIBRARY_PATH of my package but I'm not sure how?
#nix
Has anyone tried #nixMaid, a systemd-based, fresh alternative to #homeManager? I do like the concept and have had my problems with home-managers slow sequential activation.
I finally beefed up my :nix: #nix derivation for the
#gitAnnex standalone builds so the download process of the non-predictable version url is automated. It now uses git annex itself to retrieve the tarball from a matching commit in @joeyh's build repo. Funny bootstrapping situation right there 😅 But now one can just specify which version one wants and no manual steps for crafting the download url are needed anymore 🥳
Here in the #nixpkgs docs there is a vague mention that in some postgresql tests TCP connections are problematic:
But I can't seem to find any mention of some tcp connections (or protocols!?) being prohibited in the nix build sandbox 😵💫
Is there sandbox weirdness going on with
#nix fixed-output-derivations (i.e. when you specify outputHash) that some tcp connections are disallowed? I can git clone in runCommand, but down the line a 'git annex get' (which internally just downloads another file via http) fails with this weird message 'no such protocol name: tcp'. With `--option sanbox false` everything is fine, but that's no solution obviously.
https://gitlab.com/nobodyinperson/nixconfig/-/snippets/4893095
I finally tried out a new cool feature of
#gitAnnex: compute special remotes²!
Git annex is a ridiculously powerful git extension to manage large files. It remembers which file is stored on which remote. Compute special remotes now take this to another level by computing files on the fly. It's a bit like a #Makefile but built into your repository. Here¹ is a repository to try it out. It uses
#nix to set everything up.
¹https://gitlab.com/nobodyinperson/git-annex-compute-test
²https://git-annex.branchable.com/special_remotes/compute/
Next #Nix and #NixOS #Heidelberg meetup on Friday, October 17 at the Institute for Computer Science (Mathematikon). Save the date!

https://rheinneckar.events/events/f331e308-07fc-4196-8cf4-0714456a899b
@joeyh I don't understand their point. They are ranting about `make uninstall` being not ideal, LLM-generate hundreds of PRs to, what, remove the uninstall target from the Makefiles? What the hell? Then propose something like #nix as a solution. But there is no connection between the two. Nix doesn't care about `make uninstall`. Like at all. `make;make install` has always been the funny, dangerous way to try out software and definitely has its place. I don't get it 😅
The yearly
#nix / #NixOS community survey is online:
https://survey.nixos.org/759934
https://discourse.nixos.org/t/nix-community-survey-2025/68870
he choose by following some YouTube howto maybe not the most adviced choice (but he keep this way)
I call this "Linux by the north path..."
How do I install a package from my local checkout of my (modified) fork of the nixpkgs repo so that doing "nixos-rebuild switch" will overwrite these changes?
Currently the docs just say to use "nix-env -f . -iA libfoo" but so far everyone told me to NEVER NEVER EVER run "nix-env" as this will "fuck up the entire system in a way that nixos-rebuild won't fix".
Is there some documentation for which "NixOS tests in nixos/tests" I've to run when I'm trying to make a PR to nixpkgs for a single package?
At most I was able to find how to run a test but not anything in regards to which one(s).
Also as there are literally countless dedicated tests and I don't want to accidentally run one that locally rebuilds every single package the trial-and-error approach also kinda isn't fitting.
Also which of these does the CI/CD (also) run?
"To start with #Nix was one of my worst IT ideas so far."
Read about the reasons why #NixOS did not work out for me on:
Good bye NixOS, Hello #Debian (Again)!
https://karl-voit.at/2025/08/30/end-of-my-nixos/
I say bye to the great Nix community here! 👋 You were the best part of my Nix experience. 🙇
Edit: I published an extended version on 2025-08-31 with additional content and links.
I wrote a note about how to run a #Goaccess server on #NixOS: https://notes.abhinavsarkar.net/2025/goaccess-server-on-nixos
@thelinuxcast Nobody even mentioning
#SailfishOS, a good mobile Linux made by @jolla, which ticks so many of the mentioned "it's not there yet" boxes:
- well-integrated #Android subsystem (e.g install from #Fdroid or #AuroraStore)
- a battle-tested UI (that one is proprietary though)
- installable on very cheap phones
- ...
I really should've joined and talked about it. And since you mentioned it: Yes, I do in fact use the
#nix package manager on it 😉
> goes down the rabbit hole of the origin of "legacyPackages" naming
> finds no good reason (yada yada nested package attributes and nix flake show whatever) that would ever allow it to be properly addressed in #nixpkgs, the one package set most people will always have to interact with
> decides to refer to upstream #nix as legacyNix because it makes just as much sense
@alxlg @sharlatan Same here. OCI are highly reproducible, surprisingly more than I hoped for. #Nix are great... for deploying an OS and let it change through time.
I don't have too much experience for Nix, except what my coworkers used for their agency, that there is a reason why they use OCI for their software work.