The Linux Thread - The Autist's OS of Choice

[AmpleApricots]

What are your thoughts on NetBSD?

Used to have the slogan "Can run on your toaster" but that place clearly went to the linux kernel in recent years. Good compatibility to old systems, (used to know the guy who maintained NetBSD/amiga basically single-handedly, he compiled on a genuine Amiga with 68060 overclocked to 66 Mhz I think, he'd literally have that machine run for weeks) and was often their only option, nowadays I don't know anymore. The current Linux kernel still also has support for the 68k Commodore Amiga for example and you can set up a m68k kernel with drivers for it's quirks and chipset just fine, it's all still in the kernel source code. You do at least need a full-fledged 68020 for both NetBSD and Linux though, because they both need an MMU. (can't really do a modern OS with features like memory protection without MMU) I'm not sure, but *I think* Linux also at least needs an FPU and with the 68020 that's an additional, external chip. The normal 68k also has an external, optional MMU chip and I think some early Unix boxes employed that, this wouldn't work with either NetBSD nor Linux because I'm pretty sure they both require a 32 bit system at minimum. In the case of Linux, I guess it's up for people to attempt that and see if it still works which it might not. I'm pretty sure the NetBSD/amiga branch still works.

When I say *BSD I usually really talk about either FreeBSD or OpenBSD, depending on how paranoid I feel that day. I remember NetBSD had lua support in the kernel which was genuinely cool for me at that time as a big fan of the old lua. (The recent versions try too much to be a thing lua wasn't intended to be to begin with and we literal have dozens of languages who already fill that spot)

Otherwise I don't really know. My last info on the *BSDs is that FreeBSD is the one to choose if you want the best hardware compatibility for recent systems. I'd actually would need to have an actual *BSD system running to have a serious opinion and I just happen to never have the right machine for it.

Errata: The 68020 also still has the MMU as external chip, it was the 68030 where the MMU became internal in some versions, my mistake.

 

[Dread First]

What are your thoughts on NetBSD?

I know you asked the boneless peach man, but I still wanted to chime in because I too am an autist who dabbled in the BSD world. I will defer to AmpleApricots, where applicable because he's the more knowledgeable between the two of us.

FreeBSD, for all intents and purposes, is the only BSD project that the "advanced" Linux user should care about if they're sincerely considering dipping their toes into a proper Unix-like operating system. For better or worse, it is the BSD project that receives the most "mainstream" attention and it is the most widely supported of the BSD operating systems by third party software developers. While FreeBSD (and Net/Open as well) has a reputation for being more amenable to servers than desktop/workstation usage, I found that FreeBSD really wasn't that much of a chore to set up. Certainly not any more of a chore to get off the ground and running than a typical Arch/Gentoo/Debian (minimal netinstall) release is.

Around February of 2022, I began tinkering around with FreeBSD and gave brief recaps of my experiences here and here (among other posts that I'm too lazy to look for). Xbox 360 (but not Xbox One) and DualShock4 controller support was added in FreeBSD 13, and it worked pretty damn well (though not before some quick edits to text files here and there). I was even able to make good use of said controllers with Citra and MelonDS and the emulators worked at full speed with no hiccups whatsoever. GPU support was more or less "instant" once I installed Xorg along with the relevant drivers. Linux binary compatibility works flawlessly too, as the module's built off of RHEL 9. I could go on, but I think you get my point.

If you've ever enjoyed installing and getting a usable Arch/Gentoo/Debian minimal netinstall system off the ground, FreeBSD will certainly make you feel comfortable. The real difficulties with FreeBSD are far more subtle and have to do with deep-seated design philosophy differences between the Linux world and the BSD world. FreeBSD's default shells are limited to tcsh or sh (and no, it's not the Almquist shell); other shells are available via the ports collection (or pkgng nowadays), but that's not the only difference. Your home directory is located in the /usr hierarchy, as are all software packages that you install via pkgng or ports collection too. It's not too noticeable on a fresh install with only Xorg, a web browser, and 1-2 additional applications. But try using FreeBSD as your daily driver and then edit a text file you think is in /etc only to realise that it's in /usr/etc instead. You'll be groaning in frustration pretty quickly.

***

NetBSD is not worth anyone's time, unless you have a specific use case for an obscure architecture that only the NetBSD project seems to support. NetBSD has a singular focus: portability at all costs. Portability does not, however, translate into usability. The NetBSD ports collection is half the size of the FreeBSD ports collection and the binary package management tools are horrendously outdated and poorly documented (last time I tried messing with NetBSD was in 2016; take this with a grain of salt). This often means that you'll be resorting to compiling software from source on NetBSD... a lot. Software compile times are nowhere near as ludicrous as they were 10 or even 20 years ago because we have hex-core CPUs bog standard on everything nowadays, but I ain't waiting a couple of hours to compile all the Xorg bits I need, let alone Firefox.

NetBSD's obscurity also unfortunately means that getting meaningful assistance is a much more uphill endeavour. I believe that using the mailing lists is still the most efficient way to get assistance, though I could be mistaken about that. If you want to experiment with NetBSD, it's more or less an endeavour in "how much of this bullshit can I tolerate before I get fed up and install FreeBSD like I should have 4 hours ago?" Use it if you want, but I wouldn't bother.

***

OpenBSD isn't immediately "approachable" like FreeBSD is, but it rewards any seasoned Unix veteran exponential returns if they're willing to learn. OpenBSD's primary focus revolves around three things: providing best-in-class security and cryptography, code correctness and frequent audits, and remaining freely available under similar terms to the original Berkley license. I will address each one point by point.

1. Security and cryptography

Theo De Raadt et al take this point so seriously to the point of absurdity. The OpenBSD project is based in Canada, which actually has no provisions on the export of cryptographic software (a limitation in the USA that would make OpenBSD highly contentious to develop). OpenSSH was (and still is) a brainchild of the OpenBSD project, and LibreSSL was created after the first major vulnerability in OpenSSL was discovered, with the express intention of fundamentally redesigning it. The operating system revolves around "principle of least privilege," which means that the default permissions on a user account are highly limited with many redundancies in place to make sure that no retard can bork their system unintentionally. Sudo was scrapped in favour of another rewrite (doas) due to perceived vulnerabilities while making the relevant config file much easier to edit than /etc/sudoers. On the downside, you can't use more than two cores at a time because hyperthreading is apparently a security hazard which unfortunately limits OpenBSD dramatically in terms of usability. I can't comment much about their cryptography features because I never used them, but it must be reiterated: they had to be based in Canada for the sake of being able to distribute cryptographic software; they gotta be doing something, right?

2. Code correctness and frequent audits

The OpenBSD project regularly conducts frequent code audits to fix bugs, patch vulnerabilities/exploits quickly, among other such things. This is something I personally adore about the project, because they host all sorts of hackathons (and maybe a few bug bounties? I don't know for sure about that) to incentivise the community to assist developers in this endeavour. LibreSSL wasn't created because OpenSSL had one critical exploit; it's because OpenSSL was a bloated mess that needed to be cut down to size. CVE-2022-3602 and CVE-2022-3786 were two additional critical bugs similar to the one in 2014, and these ones were reported and fixed like... three weeks ago (further vindicating LibreSSL).

3. Remaining freely distributable under similar terms to the original Berkley copyright

In non-ludicrous English, this means "remaining under a lax license." While the FreeBSD project is willing to use GPL code in their base system to whatever degree, the OpenBSD project fundamentally objects to the inclusion of copyleft/GPL code (with the exception for GCC, readline, and a couple of other system libraries). Don't think they have an autistic crusade against GPL though, because they also reject any and all proprietary software and drivers (though with an exception for binary firmware).

Honourable mention: OpenBSD's documentation is also excellent. Most of it is available via man pages. man afterboot 8 is an absolute lifesaver on any fresh OpenBSD installation.

 

[AmpleApricots]

I know you asked the boneless peach man, but I still wanted to chime in because I too am an autist who dabbled in the BSD world. I will defer to AmpleApricots, where applicable because he's the more knowledgeable between the two of us.

Your post was both a lot more informed and informative than mine. I went on an tangent of a platform that's long since dead and which just doesn't seem to let me go although I only used it for comparably few years.

Yes, that's basically the thing about NetBSD, it's for obscure platforms. It also needs to be pointed out though that it has one of the better ARM supports in the *BSD world (which honestly, does not mean as much as it sounds) especially for sunxi. It is still not on the level of Linux in that regard, it's just less daunting to set up. The problem with Linux kernel obscure hardware support is - it's often there but it might or might absolutely not work. How to get it going is on you. It'll in all likelihood still compile but probably has not been tested for years by anyone. The kernel might have changed so much in other regards that running a modern kernel on e.g. aforementioned Amiga might be a fools errant for complexity reasons. That is different in NetBSD.

I'd might catch some flak for it but I'm actually kind of the border about the actual usability of OpenBSD. It's mostly from a philosophical standpoint because I think Theo's paranoia is quixotic in many ways and it makes him way too willing to just cut loose certain feature sets without warning and make the user experience worse. Yes, it's not unreasonable to believe that e.g. encryption acceleration is compromised but if so, then in all likelihood is whatever AMD/intel product you use is to begin with in various ways. There's so many ways to sneakly insert hardware backdoors into modern hardware, for example by just have them read a unique key or run a unique sequence of instructions you'd never be able to just figure out randomly, that if you already go so far to think that certain features are compromised, it's pointless to trust modern hardware, for any reason. If Theo would take his paranoia to it's logical conclusion, he'd not avoid using hardware encryption, he'd avoid using any computer that's younger than 30-40 years when they still weren't quite on the radar of these agencies and technologically just too limited to even have room for such fancy backdoors. Then again, three-letter-agencies won't use or give public knowledge of these backdoors to catch you posting gamer words on kiwifarms from your OpenBSD box. These three letter agencies are so far ahead of the curve it's pointless anyways. Look into the one time the NSA gave quite cryptic advice how to change a specific encryption algorithm, but wouldn't disclose why and how long it took to actually figure out they basically gave advice how to fix a weakness. It'll give you a hint how far ahead. Also never forget they're above the law if it comes to you. If they can't crack your fancy password, they'll just crack your kneecaps.

I wouldn't sign off on the documentation being necessarily always better either, there's some very good documentation in linux userland too, but then again, that's more of a patchwork and with the *BSDs you get a more unified experience so the comparison isn't entirely fair to begin with. This unified experience where you get a single package that includes all the bells and whistles and still has reasonable people at the helm who won't experiment with shitty, bloated, politicized trends of the day (e.g. systemd) is the most attractive thing about the BSDs in current year. This kind of screwing by certain factions you see in userland Linux hasn't quite reached the Linux kernel yet and as long as that is not the case, I can still use this kernel and just carefully pick and choose what software I put on top. If the day comes these factions also take over the kernel, it's good to have the possibility to retreat to a more sane BSD. That's what I ultimately meant with my earlier post.

I would also add that many of the BSD toolings work just fine under a linux kernel, and they're often the low-fat, better thought out versions of their often in comparison downright chaotic GNU/Linux equivalents. It just might hard to hunt them down and they might need some porting over.

 

[Trans Fat 41g]

Thanks for the input, guys.

Yeah I figured NetBSD had to go nuts, but OpenBSD in particular is admirable for sticking to their guns.

 

[Stabmaster Arson II]

Oh if we are talking about *BSD let me throw in a little tip from the last time I used it (2005 I believe, set it up as a router on an old box because I was just barebacking into a cable modem with no router back then, don't really see that now).

If you make your own scripts and stick them in $PATH (something I tend to do a lot) the shell wont find those scripts until you run the command "rehash" from a terminal. That rebuilds its list of commands/scripts in any $PATH. Kinda cool it uses a table instead of searching everything each time, faster execution time. If you don't use rehash it will rebuild it anyway on next boot, but who wants to reboot? Just a little gotcha I rarely see mentioned regarding *BSD.

 

[Dread First]

I'd might catch some flak for it but I'm actually kind of the border about the actual usability of OpenBSD. It's mostly from a philosophical standpoint because I think Theo's paranoia is quixotic in many ways and it makes him way too willing to just cut loose certain feature sets without warning and make the user experience worse. Yes, it's not unreasonable to believe that e.g. encryption acceleration is compromised but if so, then in all likelihood is whatever AMD/intel product you use is to begin with in various ways. There's so many ways to sneakly insert hardware backdoors into modern hardware, for example by just have them read a unique key or run a unique sequence of instructions you'd never be able to just figure out randomly, that if you already go so far to think that certain features are compromised, it's pointless to trust modern hardware, for any reason. If Theo would take his paranoia to it's logical conclusion, he'd not avoid using hardware encryption, he'd avoid using any computer that's younger than 30-40 years when they still weren't quite on the radar of these agencies and technologically just too limited to even have room for such fancy backdoors. Then again, three-letter-agencies won't use or give public knowledge of these backdoors to catch you posting gamer words on kiwifarms from your OpenBSD box. These three letter agencies are so far ahead of the curve it's pointless anyways. Look into the one time the NSA gave quite cryptic advice how to change a specific encryption algorithm, but wouldn't disclose why and how long it took to actually figure out they basically gave advice how to fix a weakness. It'll give you a hint how far ahead. Also never forget they're above the law if it comes to you. If they can't crack your fancy password, they'll just crack your kneecaps.

I largely agree with what you're saying, though I would argue you're being a bit too dismissive of it. OpenBSD's sane defaults make it far more suitable for network server applications than FreeBSD is. I'm not a huge fan of Mental Outlaw, but he made some pretty good videos on OpenBSD if you're curious as to how it runs in a dedicated server space.

 

[std::string]

If they can't crack your fancy password, they'll just crack your kneecaps.

Agents taking bets on how many times you get to ride on the waterboard before you start yelling out your boot password.

Shit I have a Linux book from the late 90s that talks about TEMPEST. I mean look at this fucking shit (PDF link.)

If you got state actors after you it's not going to matter if you're running OpenBSD.

 

[Dread First]

If you got state actors after you it's not going to matter if you're running OpenBSD.

Edward Snowden is the only man on the planet who can evade state actors on some measure, and even then, that's only because he was a glownigger a decade ago.

If you sit down and you watch all of the Joe Rogan interviews, Snowden goes into excruciating detail about how everything is sending data at all times along with the steps he takes to mitigate everything he can. I'm talking about de-soldering certain chips on whatever device he's setting up, isolating mobile devices in specially enclosed rooms that dampen radio signals, you get my point.

How this man hasn't gone insane in the decade since he leaked NSA PRISM to the world is beyond me.

 

[Ahriman]

Alpine 3.17.0 released | Alpine Linux

Alpine Linux

www.alpinelinux.org

The comfiest distro out there has a new release. Alpine + LXQt seems to be a pretty nice and unbeatable combo these days.

On a lighter side of things...

 

[AmpleApricots]

I largely agree with what you're saying, though I would argue you're being a bit too dismissive of it. OpenBSD's sane defaults make it far more suitable for network server applications than FreeBSD is. I'm not a huge fan of Mental Outlaw, but he made some pretty good videos on OpenBSD if you're curious as to how it runs in a dedicated server space.

I can believe that, in server environments you usually also have other requirements. I got an older Skylake convertible to replace my very cheap netbook. (what mostly annoys me is laptop keyboards. I hate them with a passion of a thousand suns and have never seen a good one. Yes I include old thinkpad keyboards too. the idea behind a convertible/tablet was that I can pretend the keyboard part never existed and use a small, proper keyboard with it) And I just realized (and a quick google confirmed) that it's hardware should be supported by both Free- and OpenBSD as it's quite old by now. Maybe I need to finally give them a spin in current year so I can actually know what the current deal is. I've got so used to basically do everything by myself, I don't really know what it is like to run "complete packages" like this anymore.

The comfiest distro out there has a new release. Alpine + LXQt seems to be a pretty nice and unbeatable combo these days.

If I didn't use something homegrown I'd probably use something like alpine nowadays. (Gentoo kinda disgusts me at this point) I use alpine for my containers sometimes. Love it, even though their musl-thing sometimes breaks my balls with proprietary software.

 

[Dread First]

I can believe that, in server environments you usually also have other requirements. I got an older Skylake convertible to replace my very cheap netbook. (what mostly annoys me is laptop keyboards. I hate them with a passion of a thousand suns and have never seen a good one. Yes I include old thinkpad keyboards too. the idea behind a convertible/tablet was that I can pretend the keyboard part never existed and use a small, proper keyboard with it) And I just realized (and a quick google confirmed) that it's hardware should be supported by both Free- and OpenBSD as it's quite old by now. Maybe I need to finally give them a spin in current year so I can actually know what the current deal is. I've got so used to basically do everything by myself, I don't really know what it is like to run "complete packages" like this anymore.

OpenBSD allegedly supports AMDGPU and Ryzen, though the rule of thumb nowadays is "If an OpenBSD developer uses that hardware for their personal machine, the support will be readily available." I haven't personally tested this yet, because I'm quite comfortable with Rocky Linux 9 on my spare hard drive and I don't feel like nuking it again to satisfy an impulse project.

 

[glicko45]

Which BSD is the best for everyday use in your opinion?
*FreeBSD and derivatives*
*OpenBSD and derivatives*

 

[Sonichu-choo trainwreck]

OK you autistic white supremacist nazi transphobe incels

Since it's all the black friday (obviously racist), cyber monday and all that shit...
what would be a good little cheapie laptop ($300?) to use as a full-commit platform to try out linux and/or BSD?

I just figured cheap so no tears/hesitation at fully committing it to a new OS/ecosystem
and by "good" I think one of the things I'm hoping for is decent driver support for that HW - since I'm just looking at those OSs, I'm hoping not to get bogged down in "Oh, it's got THAT touchpad in it"* driver toubleshooting so I can just make a go at learning the OS.


*actual problem I ran into abt 5 years back trying

 

[seri0us]

OK you autistic white supremacist nazi transphobe incels

Since it's all the black friday (obviously racist), cyber monday and all that shit...
what would be a good little cheapie laptop ($300?) to use as a full-commit platform to try out linux and/or BSD?

I just figured cheap so no tears/hesitation at fully committing it to a new OS/ecosystem
and by "good" I think one of the things I'm hoping for is decent driver support for that HW - since I'm just looking at those OSs, I'm hoping not to get bogged down in "Oh, it's got THAT touchpad in it" driver toubleshooting so I can just make a go at learning the OS.

HP Elitebook 9480M

 

[Sperg Coalition]

View attachment 3928968

Alpine 3.17.0 released | Alpine Linux

Alpine Linux

www.alpinelinux.org

The comfiest distro out there has a new release. Alpine + LXQt seems to be a pretty nice and unbeatable combo these days.

On a lighter side of things...

I don't know if Alpine is suited for desktop usage, since it uses the musl c library which can have some problems with proprietary applications such as Steam (granted, I can create a chroot with glibc and use Steam from there, but still).

Snap is shit compared to a distro-native package, I dont't understand why some people like it tbh

 

[AmpleApricots]

I often used and abused Alpine as container basis and never really looked at it past that. Seeing it mentioned here made me actually look closer and it's actually really decent for a binary distribution if you want to have some more fine control. What I like most is that it doesn't make any assumptions or forces you to use it in a certain way. Beyond the kernel and busybox and the package manager itself there aren't really any hard dependencies. There's no default shell, there isn't even *really* a default init if you start with the minimal rootfs and just build your system from there. Even the more complex packages are compiled with the most minimal possible dependencies in mind. (e.g. alsa instead of pulseaudio/pipewire/whatever is the new big thing now) Hell, even the man-pages to the stuff you can install are optional, additional packages. Quite good actually. It could be daunting for less experienced user to set up though. If you're used to source-based, rolling distros but want to try something more stable, convenient and lower maintenance while not giving up too much control to the distro tranny jannies, alpine might actually be worth it.

For musl there are apparently compatibility shims now too, otherwise there's always the way of just using something else in a chroot/user namespace, as mentioned.

 

[Another Char Clone]

Snap is shit compared to a distro-native package, I dont't understand why some people like it tbh

Download thing, run thing. Its the Windows experience

 

[Cum Elemental]

OK you autistic white supremacist nazi transphobe incels

Since it's all the black friday (obviously racist), cyber monday and all that shit...
what would be a good little cheapie laptop ($300?) to use as a full-commit platform to try out linux and/or BSD?

I just figured cheap so no tears/hesitation at fully committing it to a new OS/ecosystem
and by "good" I think one of the things I'm hoping for is decent driver support for that HW - since I'm just looking at those OSs, I'm hoping not to get bogged down in "Oh, it's got THAT touchpad in it"* driver toubleshooting so I can just make a go at learning the OS.


*actual problem I ran into abt 5 years back trying

Dell Latitudes are solid. Well built and they don't have the ThinkPad meme tax. Most Latitudes had Ubuntu as a factory OS choice. They also tend to have Atheros WiFi cards (FOSS drivers) which makes the install easier.

Latitude 7280 and 7290 are coming up on End of Life/going out of warranty so businesses are starting to liquidate them, which means they're showing up on the secondary market for cheap.

If you can find a cheap ThinkPad x220 or x230 they're pretty decent and have known good Linux compatability. Prices are inflated for the hardware that you get. They're starting to show their age though and get fucked by VP9 streams above 720p. Don't buy a ThinkPad made after 2016, Lenovo stopped giving a fuck after the xx60 series.

Cant speak for HP EliteBooks or Toshiba Portege as i've never worked on them.

TL;DR Get an ex-lease business laptop, preferably a Latitude.

 

[Celebrate Nite]

Download thing, run thing. Its the Windows experience

so is appimage, and you don't need a package manager for those.

 

[Another Char Clone]

Then its the Ubuntu branding, probably.