Live data from Hacker News

Why Gentoo?

blogs.gentoo.org

21–30 of 85 posts

Re: Why Gentoo?

#21
post #15

I've been using Gentoo since 2003, and I don't really plan on changing any time soon... > We don’t go out of our way to tell you how to use your system I'm not really sure I believe this, because the default profiles do exactly that. One day you'll be happy with your pam config, then bam, package updates and you're forced into passwdqc and faillock.

> One day you'll be happy with your pam config, then bam, package updates and you're forced into passwdqc and faillock. Are you saying you've had your PAM config broken by a package update when you didn't update anything in /etc, or that you had your PAM config broken by a package update where you blindly accepted changes to stuff in /etc?

This a trap! Don’t answer!

Re: Why Gentoo?

#23

Gentoo is customizable. But really customizable. That's my #1 reason for using it.

Gentoo let me install mpg123 on a headless box without bringing in X, which is what pissed me off about RedHat at the time.

Re: Why Gentoo?

#24
I agree 100% with this article. However I switched to NixOS several years ago.

On the axes I most care about, NixOS is better than Gentoo. In particular, managing configurations in NixOS is really a breeze. No more merging diffs of random files in /etc.

On one hand, you could say that Nix has more magic than Gentoo, but on the other hand, the online nix option search links directly to the source code implementing the option.

Gentoo wins on documentation and supporting more than one init system. It probably also wins on security; I haven't dug recently, but NixOS doesn't have a great story for e.g. Mandatory Access Control. Also the nix store is world-readable, so it is much easier than it should be to accidentally spill your secrets to the entire system.

Re: Why Gentoo?

#25
post #7

I think it's time to use an LLM to rewrite portage in C. Might be one of my next projects. Should not take long and the result will be easy to deterministically test side by side.

Switch to using a git backend and portage will fly fast enough.

  tank ~ # time emerge --sync
   * Using keys from /usr/share/openpgp-keys/gentoo-release.asc
  emerge --sync  3.18s user 5.15s system 67% cpu 12.324 total
https://wiki.gentoo.org/wiki/Portage_with_Git

Re: Why Gentoo?

#26
post #3

I didn't realize how strict they were against LLMs. Codex has really helped my fix and tighten up my AUR PKGBUILDs. I was thinking about trying Gentoo in the future, but not being able to contribute because I use LLMs in my workflow sucks.

If you're so good at it that they can't tell, does it matter?

Re: Why Gentoo?

#28
post #18

What kinda baby would come from mashing together gentoo, void, freebsd, and nix?

nixos is already, from a certain perspective, an exotic take on gentoo. take away cache.nixos.org and it becomes a source-based distro. it's not even too out-of-the-question to do traditionally gentoo stuff like globally set -funroll-loops -O3 and rebuild your whole system, but the operational benefit of staying standard is getting to use binary caches since reproducibility lets them substitute transparently for local builds in the standard environment with the standard settings. one thing nixos lacks is an elegant analogue to gentoo USE flags; "tell package A to use optional package B" has like two different conventions in nixpkgs for individual packages, alongside bespoke things like configuring everything to use cuda or not. furthermore, pkgsrc, gentoo prefix, and nix are three of a kind in terms of "third-party package manager on a non-native OS"

personally, i used gentoo throughout most of my teenage years, and now use nixos viewing it as a successor, since you can screw around using the thing as a meta-distribution, but roll back when you hose your system.

gentoo freebsd existed for a while for funsies but nobody cared and now it's dead, same with gentoo openbsd. each was portage integrated a little more intentionally on top of a respective bsd than just "bung it in /usr/local/bin and call it a day". practically speaking, they were implementations of gnu/k*bsd.

nixbsd similarly exists (nix deeply integrated into freebsd), with heroic efforts made to transfer nixos's abstraction over init systems for system-wide configuration; last commit was two months ago so i guess not dead yet?

you can run nix on gentoo, or gentoo prefix on nixos with nix-ld turned on if you feel like it

void's claim to fame with xbps is a bunch of sandboxing and bind mounts for build environments. nix already has to sandbox just to get off the ground wrt reproducibility.

void, gentoo, and nixos all offer some degree of libc freedom. on void it's first-class, on gentoo it nearly is, and on nixos it's here be dragons. libc freedom doesnt really jive with using a bsd as a base; vertical integration is kind of one of the contemporary selling points.

so to answer your question, the result would be nixbsd in an alternate timeline where nixpkgs's analogues to USE flags/eselect/profiles are not hot garbage (i say this with love as a user), and void would fail to exert a phenotype.

Re: Why Gentoo?

#29
post #8

Never used Gentoo, but what is its advantages over, say, Guix? I don't actually use guix or nix as a daily driver, although one year ago I did install Guix on a separate partition of my desktop PC and I use nix via Home manager to install various cli-based tools.

The main selling point is that you compile out any features you are not using, on a global level. Want to remove PulseAudio? Add -pulseaudio to the USE flags and rebuild. Prefer GnuTLS over OpenSSL? Declare it in make.conf and packages that support either will only use your choice.

That's unique to the Gentoo model and gives a fine-tuned system with virtually zero bloat. Other distros inevitably pull in dependencies you don't need because they are linked at build-time in case someone relies on it; choices like crypto backend are made for you by the packager.

Guix lets you tweak things on a per-package level, but currently offers no global feature flags. You can do wild stuff like replace OpenSSL with LibreSSL everywhere, but only as long as they are API- or ABI-compatible.

Re: Why Gentoo?

#30
post #6

Earlier quoted context omitted.

I know for a fact that some stuff merged into Gentoo is already LLM modified in upstream so their stance is dumb and not even logically consistent.

It's pretty consistent. If you check out something from a Gentoo-official repo, it's going to have been created by humans without the use of LLM tooling. LLM tooling is used for the kernel, systemd, and a litany of other packages -- trying to run a "non-LLM OS" is more like a fork of the whole ecosystem than a distribution. I can appreciate the concept that Gentoo is placing a high value on human curation . I used to…

Been using Gentoo for over 20 years and have accepted PRs to the repo.

I don't get any value from this "no LLMs" stance. In fact it's quite the opposite.

I watch extremely famous C devs at work using LLMs all day everyday working on stuff that is merged into Gentoo all the time. You guys are shooting yourselves in the foot and just wasting everyone's time and it's exhausting.

Post reply on HN