1. 23
  1. 11

    Do note this is just a maintenance release for the old version.

    The current release is 9.0, and that’s what’s recommended for new installs.

    1. 4

      Anyone here running NetBSD as their primary desktop? I’d love to hear your stories :-) Can you share some of what you like about it? The small moments where you stop and think, that’s why I run it!

      1. 11

        I use it (have for a few years). I really enjoy how quickly I can go from “name of a binary I use” to the source code that created it. Applies to both base and packaging.

        It’s also a really nice simple system that I understand fully, which is always fun. It brings me a lot of confidence in doing insane things like live-swapping libc to test out a patch.

        My use case has been very FOSS-heavy for many years, so I don’t miss much. I did quit some video games in my transition, but it felt like a good decision (the particular way I was playing had negative effects on me).

        During a few days I felt like I want to work from my NetBSD machine, I was using it as a SSH terminal to a remote Linux box, and using my Android phone for video calls.

        My graphical environment is actually harder to reproduce on the user-friendly linuxes, since I use things like dwm, which are configured by patching code & rebuilding.

        I’ve been a developer for NetBSD for the past few years and it has encouraged me to do crazy things I would never have otherwise attempted, like a lot of toolchain and driver work. It has given me justified self-confidence on the professional front. It’s a nice community and I recommend it.

        1. 3

          I’ve been a developer for NetBSD for the past few years

          Get a hat! https://lobste.rs/hats

        2. 8

          I don’t anymore (it simply doesn’t support things that I need for my job), but I did for a long time years ago.

          • pkgsrc is great.
          • The whole rc system is really nice.
          • The work that went into making it very portable also resulted in a clean and understandable code base for the code system components.

          NetBSD also seems to be doing the most moving-the-ball-forward of the BSDs of late. OpenBSD is trying to be the most secure which is great, but that encourages not doing things like putting Lua in kernelspace. :) The rump kernel project and various other bits of neat stuff is great.

          1. 5

            NetBSD also seems to be doing the most moving-the-ball-forward of the BSDs of late

            well.. jmcneill@ is doing incredible development with all the things Arm and embedded (RPi4 Ethernet most recently, AWS a1.metal, display output on lots of SoCs…), for sure.

            But in general, FreeBSD is rolling lots of balls forward, I would say more than the others, but I’m biased of course :)

            The rump kernel project

            unfortunately rumprun doesn’t seem to be actively developed anymore..

            1. 4

              Sorry, I didn’t mean to discount others’ hard work. I meant something more along the lines of “willing to do more researchy/experimental things.”

              Then again, I may be wrong on that front too.

              1. 1

                That HID work looks promising. It would be good if it enabled N-key rollover in keyboards to allow things like Plover and my unreleased stenography engine to work on FreeBSD.

                1. 1

                  N-key rollover in keyboards

                  It does, I just checked — my USB keyboard went from 6-key to >20-key rollover.

                  I’m actually kinda surprised it didn’t work with the stock ukbd — I thought the keyboard pretended to be a hub with multiple keyboards, but no, it’s multiple HID endpoints on one device.

                  1. 1

                    Excellent. Thanks for checking.

              2. 1

                What is good about putting Lua in kernelspace?

                1. 6

                  So far they’ve used it for scriptable CPU scaling and packet filtering, which is cool. The paper where they discuss the implementation points out that their packet filtering does not introduce significant performance overhead and does filtering stuff that would be impossible with existing solutions.

                  The idea of having a memory-safe scripting in the kernel instead of more C seems like it could be a good thing for when things need to be more dynamically configured. If they had gone with userspace controllers, they’d need to poke more holes into the userspace-kernelspace barrier. This way, a lot of functionality can be exposed through the single “run a Lua program” thing.

                  Most importantly, though (and IMHO), it doesn’t matter whether it’s good or not. We can’t know if it’s good or if it’s useful until we try it, and I’m glad that the NetBSD folks were willing to give it a try. If they fail, no harm no foul, but if they succeed, they’ve pushed the state of the art a little bit forward.

                  1. 2

                    Yeah, I didn’t know about this either. Found slides on it. Looks pretty cool. One motivation, prototyping or debugging drivers, is similar to Rump kernel.

                2. 6

                  Anyone here running NetBSD as their primary desktop?

                  Yes, I am.

                  I had an old netbook idle at home and decided to make that my primary computer (opposed to the family computer). Have in mind, though, I primarily use my computer to write code and I am a quite new NetBSD user.

                  In a nutshell:

                  Pros:

                  • it is fast (faster than my previously installed Arch Linux in that same machine);
                  • it is “clutter free”—it has (almost) zero default services, and I know and understand all the process running in the operating system;
                  • it is documentation (both the “The Guide” and the man pages) is the most complete and “human friendly” I have ever seen—take a look on the man afterboot to understand what I mean;
                  • there is a level of symbiosis between the kernel and userland is incredible (for a person used to Linux);
                  • pkgsrc, although I just use the binary distribution (pkg_add and pkgin) so far;
                  • it is the first *nix system I feel comfortable to not have a window manager (I have i3 for the rare occasions I need to use Firefox, though);
                  • it seems (by what I see in their official blog) to be actively improving security (e.g., kernel fuzzing efforts) and compatibility with LLVM, which I find interesting.

                  Cons:

                  (I sincerely cannot say: everything seems to be working—I haven’t tested the Bluetooth—and I am very glad with it! Let’s try something for the sake of argument.)

                  • it requires some experience with *nix to install and configure it, although not much—I doubt my parents would use it instead of Windows;
                  • it forces you to read more documentation;
                  • pkgsrc binaries are updated less frequently (every quarter of the year) than in the apt repository;
                    • I have yet to learn how to use the pkgsrc properly, and it requires some documentation reading;
                    • even when I learn to use ir properly, I will need a lot of luck, patience, time and electricity compiling Firefox in a netbook.
                  • sailor is not there yet, while FreeBSD has Jails.

                  Why NetBSD?

                  All in all, I thought: why not?

                  1. 2

                    @guiraldelli, thanks a lot for your detailed reply and links! I’m going over them all. :-)

                  2. 4

                    Not as primary, but I’m running it on two oldish computers, an Amiga 1200 (with a 030 and fpu) and a PC with one of the first Athlon CPUs (Slot A).

                    I love it. The kernel and userspace are non-bloated (relative to Linux), the manpages are excellent and the source code is well-structured and pleasant to read.