I run Void on most servers and clusters, yes. I also run Gentoo on my main desktop (a Gentoo distro I've continually updated since 2012 based on a make.conf from 2004) and it runs RAID+LUKS+btrfs (and I'm in the process of moving it over to zfs):
...all that being said, Gentoo can be a total bitch of a distro. I've been running it so long I've just gotten really use to it and have a lot of custom stuff in my local Gentoo overlay. That being said, the concept of overlays is really handy; something that Void is NOT suited for.
The Void devs absolutely refuse to accept templates for Hyprland (closing all issues with zero explanation). This person made a custom repository. So you can make custom repos, but you gotta copy your "overlay" into the standard package management repo in order to build custom packages. 3rd party packages are not a first class idea on Void:
I've been experimenting with using nix to build containers. That way I can rebuild easily by changing the commit ID and easily roll back/forward to a different version of the package tree. In theory, you should be able to implement a nix-type immutable OS with any distribution that uses version control for the entire package repository.
You just create a build system that builds all relevant packages for a given commit; build a set once a day and BAM! Immutable system with rollback/forward (although no cool `/nix/store` and you can't mix and match different versions; unless you use it to build a chunk of containers).
Honestly nix should have been a source only distro, with the compiling happening locally. The amount of packages they build and store really isn't sustainable long term (same kinda goes from npm, pypi, gems, etc ... wondering when the day comes a lot of them start to purge old packages).
@mischievoustomato@prettygood >Couldn't you remain safe if you update monthly or weekly? That's what I with arch and beyond upstream devs being retarded, all goes well. From my own experience (at least with Arch), with a rolling release there's always a degree of uncertainty especially for cutting/bleeding edge. It's not if you'll get shit breaking because libgayniggers.so.6 was not found, it's when. You can't easily predict behaviour. On a desktop I was more lenient towards doing a quick downgrade on occasion, on server I have zero tolerance for that. I need shit to be stable, and above all else that means predictability.
Addendum: I hear Void places a higher emphasis on stability despite being rolling release. Also xbps-src looks like what I want. I know @djsumdog runs Void even on cluster servers so Imma tag him for input.
>how so? If there is a piece of software for Linux, 99.99% of the time it has a .deb package. Even if it's not in repos I can dpkg -i and have absolute confidence it will run. I can't say the same for Gentoo but I'm not as familiar with it.
>i don't like python, but i've heard tha portage just works well. maybe @prettygood can give some insight would appreciate it.
@mischievoustomato I know Gentoo does but I prefer versioned releases for server OS. Plus it's not as well-supported, the package manager is in fucking Python and when I tried to install I got lost on encrypting my drive. IMO every distro should at least have a script or an option to automatically set up LUKS.
NGL I wish all distros had a ports system neatly integrated into the package manager. I just want to autistically compile some packages with -march optimizations and keep them up to date using the same package manager as the one for my binary packages.