FreeBSD rather than Linux?

[DiTBho]

Either way, FreeBSD is a lucky OS with an exceptional x11 support, so that's not a problem 

[nightfire]

From my experience with FreeBSD from 3.2 on: FreeBSD back then was a great and very stable OS for servers, and minimalistic workstations.
My homebuild NAS still runs on it, and I am in the progress to recycle my old PC as a Unix Workstation, also used hopefully in the future for some video capturing, if I get the whole blackmagic hardware to work.
I also used FreeBSD on my notebook at my last employer, where security was a concern- basically only with the most important role to work as a screen multiplexer for serveral xterms.

The main differences between FreeBSD and Linux (resp. GNU operating system with linux kernel...) is in the support of new, fancy and special hardware. Within the penguin community there are much more ressources to come up with drivers for exotic stuff, and/or some commercial vendors take part in providing ressources to get this stuff running.
Therefore the driver support under Linux is usually a bit faster to provide drivers and a bit more versatile.
Also most software that runs under Linux can be used with FreeBSD, but in reality it is sometimes a big hassle to compile them with all dependencies yourself.
On the other hand: The ports system is basically a curated software repository with lots of stuff to choose from, which usually works quite well, including updating the whole enchilada without breaking dependencies.

ATtention: With notebooks it is very vital to check hardware support first, especially Windows-oriented companies like HP are know for some picky hardware.

[Normamd]

FreeBSD is a monolithic kernel with it's primary interest being stability with security as a secondary but important concern. Linux is also classified as a monolithic kernel, however it's much more modular to the point I really consider it more of a hybrid. There's pros and cons to both approaches.

[Ampera]

I used FreeBSD for a good few years, and picked it up around the same time I learned to use Linux. In that time, I found it to be a great operating system, with a lot of good ideas, and simple implementations for the user. It ended up doing some good work for me, until I realized I needed to do more with my computers.

I've used FreeBSD as a workstation operating system, and it really doesn't bring anything to the table in that regard over Linux. Speaking practically, GPU drivers tend to be a lot more labored, there will be things you didn't even realize were Linux only until you need them, and even when you do get everything put together, it'll just end up feeling the exact same as a Linux workstation system. The only advantage I could see, is the FreeBSD model of stable base system combined with rolling ports is /excellent/ for a workstation user, as it gives you recent software, while almost never putting you into an unbootable system due to updating your packages.

Therefore, FreeBSD was largely a server operating system for me. I can proudly say that I managed to run, as one of my first real internet projects, a 80-100 million connection/mo BEMP (that's BSD, Nginx, MariaDB, and PHP) cluster, using nothing but FreeBSD. ZFS out of the box was excellent, and stable base + modern ports wasn't a bad mix here either.

However, sitting here, having not too long ago moved to Debian for all my servers, I can say I would have absolutely not done it the same way. ZFS on Linux is stable, and despite being a minor pain to get going, works just as well as FreeBSD. A lot of the software is far more optimized on Linux, with my now LEMP stacks having far faster response times, and far less CPU load for the same workloads. I found across the board, especially in compute intensive Java workloads, that Linux just ran faster. The Debian project is a marvel of open source software, and despite FreeBSD being famous for shipping an operating system not a kernel, I still find Debian to be the more polished option.

And then there is this: https://vez.mrsk.me/freebsd-defaults.html which for me was the final nail in the coffin. A lot of the time, even if we think we know what we're doing, we're ultimately reliant on the people who write the software we use to know what they're doing, and make responsible choices. While I'm sure for a lot of what's said on this page has more than one side, from the digging I've done, I'm honestly disappointed with what seems to be a project more interested in chilling in the 90's than in writing a safe, secure operating system.

I'm no expert, this is just my personal experience, and I'm sure half of what I've said is wrong. Just take it from my experience that just because it looks cool, and just so happens to be to Linux as Linux is/was to Windows, it doesn't mean that it's actually worth using. For servers, I use Debian, and for my workstations (primarily for convenience in installing weird software, the AUR is awesome), I use archlinux.

[RoGeorge]

And then there is this: https://vez.mrsk.me/freebsd-defaults.html which for me was the final nail in the coffin.

I've took a look and can not comment about the security problems (because I don't know enough to audit such things), or if the issues were addressed meanwhile or not.  However, I did saw something easy to check

It seems that anyone with commit access can dump in whatever they want (or, more often, what their parent company wants) without any communication with other FreeBSD developers or any code review. This "commit-then-discuss" culture usually leads to lots of drama, long arguments on the mailing lists, security problems and developers leaving the project over political commits. Most of this could probably be avoided if some review had taken place first.

Source:  https://vez.mrsk.me/freebsd-defaults.html

The submitted change was for changing "Master Boot Record" into "Main Boot Record", sponsored by former Mellanox Technologies:

Mellanox Technologies Ltd. (Hebrew: מלאנוקס טכנולוגיות בע"מ) was an Israeli-American multinational supplier of computer networking products based on InfiniBand and Ethernet technology. Mellanox offered adapters, switches, software, cables and silicon for markets including high-performance computing, data centers, cloud computing, computer data storage and financial services.[3]

On March 11, 2019, Nvidia announced its intent to acquire the company for $6.9 billion.[4][5] Other companies willing to acquire Mellanox were Intel, Xilinx and Microsoft.[6] The deal closed on April 27, 2020, with approval from the EU, U.S. and Chinese antitrust authorities.[7]

Source:   https://en.wikipedia.org/wiki/Mellanox_Technologies

However, I've checked in my current FreeBSD install, and inside the file '/usr/src/contrib/file/magic/Magdir/filesystems' the text reads "Master Boot Record". 

[Ampera]

I hardly consider that to be the biggest problem on that list.

[alm]

So next topic "OpenBSD rather than FreeBSD"? 

[Ampera]

So next topic "OpenBSD rather than FreeBSD"? 

There's a lot to respect about OpenBSD, not just for its work to create a secure, well written operating system, but also as the home of projects like OpenSSH. It and NetBSD are quite well ported, sometimes with better oddball platform support than most Linux distributions. That being said, its partitioning scheme is archaic, the best filesystem you get is UFS, and you're not going to be running Steam games on it any time soon.

[SiliconWizard]

That was interesting to get some feedback/views on FreeBSD. At this point, I am now convinced I'm not going to bother with it for workstation use. Unless you pay me.
But as some of us said, it's always good to have more options.

One of my points, beside curiosity, was that it was possibly easier to distribute software on FreeBSD (due to its integrated nature) rather than on "Linux" which is a moving target with hundreds of different distributions and different versions of the same distributions. Or, as many vendors do, you only "target" (officially support) Red Hat or the like - which is not something I'm very fond of.

[DiTBho]

Microkernels like newos are better for academic reasons.
Well, I like them, but ... porting haiku (based on newos) to riscv is a very hard job.
It s harder with microkernels that with monolithic ones.

[Nominal Animal]

One of my points, beside curiosity, was that it was possibly easier to distribute software on FreeBSD (due to its integrated nature) rather than on "Linux" which is a moving target with hundreds of different distributions and different versions of the same distributions.

Perhaps it is time for a guide on how to make portable Linux binaries, then?  It is doable, although it is somewhat complicated.

(The reason I haven't yet put anything up on my home page is that I'm particularly weak/sensitive to the situation when I show how something can be done portably, fixing issues, having the result be easily maintained and very robust... and being completely ignored because I'm not popular enough.)

(I do not get a kick out of showing others how to do better. I only get a kick out of helping others to actually do better, seeing them do things neither of us alone could have done.)

[SiliconWizard]

One of my points, beside curiosity, was that it was possibly easier to distribute software on FreeBSD (due to its integrated nature) rather than on "Linux" which is a moving target with hundreds of different distributions and different versions of the same distributions.

Perhaps it is time for a guide on how to make portable Linux binaries, then?  It is doable, although it is somewhat complicated.

If portability is really guaranteed, then yes. Certainly.

[Nominal Animal]

One of my points, beside curiosity, was that it was possibly easier to distribute software on FreeBSD (due to its integrated nature) rather than on "Linux" which is a moving target with hundreds of different distributions and different versions of the same distributions.

Perhaps it is time for a guide on how to make portable Linux binaries, then?  It is doable, although it is somewhat complicated.

If portability is really guaranteed, then yes. Certainly.

Only to a limit, of course; specifically, library dependecies.  Kernel interfaces are stable enough to not worry about.  PulseAudio and SystemD are a bitch (because they are only compatible with themselves, and are not really even backwards compatible with themselves), not sure about integration with those (except via widget toolkits like Qt, Gtk, FTLTK etc).  Gnome-dependent applications on gnomeless distributions can also be a bit problematic, due to all the crap it spews to user home directory (.config/, .cache/, .local/).

It is a matter of bringing along the set of required libraries, and using a launcher script and a directory full of symlinks to resolve the library dependencies.  Whenever the host system has a newer set of compatible libraries, those are used; otherwise the packaged-with libraries are used.  Nice thing about this is that while the actual binaries and scripts should be system-wide and owned by root (for security; we do not want applications to generally be able to modify themselves), each user can make their own launch setup with minimal cost (shell script and symlinks).

(For services that run as a dedicated user account, the work is easier, because the spewage to their home directory will not conflict with anything else that user account would be doing.)

Can you think of a suitably annoying to package open source program I could use as a test and demonstration?  Being open source, we could ask other members here to test it on their Linux machines.  Graphical would probably be best, but small enough (in binary form) to not tax internet connections for those who'd like to test.  I do have plenty of experience with Debian and RPM packaging and Linux integration in general, but do need to get up to date with Pacman (Arch Linux), so it'll take me some time.

[DiTBho]

The reason I haven't yet put anything up on my home page is that I'm particularly weak/sensitive to the situation when I show how something can be done portably, fixing issues, having the result be easily maintained and very robust... and being completely ignored because I'm not popular enough.

Even great filesystems got completely ignored.
People wishes ... Is what rules.

[SiliconWizard]

Can you think of a suitably annoying to package open source program I could use as a test and demonstration?

Well, I don't know... Something using GTK3 for instance. Maybe GTKWave.

[Zucca]

I'm late to this party.
Interesting discussion.
For Server I am a big FreeBSD guy. I have even my Octoprint on my dedicated FreeBSD client.
For workstation my wet dream would be Arch Linux, but I never tried it and I do not know if I am driving against a wall.

My target: get rid of Windows, I want my work horse(s) stations to be on Linux. Why? Because it's my PC not theirs, I want 100% control....

[SiliconWizard]

For workstation my wet dream would be Arch Linux, but I never tried it and I do not know if I am driving against a wall.

I've been using Arch as my sole distro for years. Currently all my Linux installs are with Arch (laptop, a headless box, and even RPi's with archlinuxarm.) The only Linux thing not on Arch that I still run is a NAS I installed years ago with Centos and that still runs fine (but these days I would NOT use Centos for anything new.) Of course Arch requires some solid prior experience with Linux, preferably. But if you are in that case, go for it. The wiki is well made for the most part. If you are less experienced, I'd suggest starting with a more "beginner"-friendly distro. If you'd want to go Arch eventually, you can try Manjaro. It's based on Arch but much easier to install and configure than raw Arch.

My target: get rid of Windows, I want my work horse(s) stations to be on Linux. Why? Because it's my PC not theirs, I want 100% control....

Despite the above, I still run Win (7) on my main workstation for various reasons, which are definitely not because I don't know how to switch. The Win days are probably now being counted though.

[Nominal Animal]

OSes are just tools.  I for one am happy that completely different OSes exist, and do not believe for a second that using Windows is not indicative of anything, AT ALL.

What does grind my gears is the talk about how Linux or whatever should be more like Windows, or that there being so many Linux variants out there is a bad thing.  The fossie OSes are developed in a completely different ecosystem, one where being an user has zero or negative value, and only being a contributor –– be that money, time, effort, or all three –– has positive value.  Even if you sell Linux software and make big bucks, that still applies!  The wide variety and variance ensures the solution phase space is explored, and we have better distributions due to the competition.  Those are the reasons why any kind of argument about "Linux won't X the Y market, until ..." is so aggravating.  It's like telling beekeepers that unless they start using electric vehicles, the bunnies on the neighboring fields will keep eating the daisies.  Y'know, irrelevant babble.

In other words, I don't like it when others perceive me as opposing Windows, because I do not.  I oppose expanding Windows-isms like systemd to other OSes, but I do not think there is anything wrong in using Windows if it works for you: it's just an useful tool, nothing more, nothing less.

In fact, I rather feel similarly wrt. FreeBSD and Linux.  There are POSIX.1 system interfaces that are useful even in BSDs, and I don't consider them Linuxisms (because POSIX.1 originated from Single Unix Specification), but expecting Linuxisms like /proc and /sys pseudofilesystems to be available on non-Linux systems is kinda evil.  I do use those on software that I know is difficult to port to *BSDs anyway, but I do feel a bit guilty then...  Much better is to put the OS-specific details behind a common operational API, and use build-time overridable preprocessor macros (look here for autodetection by default) to choose the one for the target OS.

Some things that look like Linuxisms, like using MAP_ANONYMOUS, are actually natively supported on FreeBSD, too, even though man 2 mmap on Linux kinda claims they're "Linux-specific".  (MAP_NORESERVE, on the other hand, was defined for a while on FreeBSD, but never had any effect.)

[Zucca]

I've been using Arch as my sole distro for years...

Call me crazy, but I read the entire FreeBSD manual and it was so nice. Definitely an eyes opening experience, after years of DOS and Windows- I have no fear of FreeBSD anymore.
I even built my custom kernel.... That said I am not an expert, FreeBSD is so immense.... and I have limited time to test things out.

I am trying to repeat the same with Linux, Arch seems the best technical documented out there. I will setup a virtual machine and slap Arch on it, after I read through the Arch wiki....

Correct me if I am wrong, but once you can deal with Arch you can jump in any other distro with ease.

Knowledge is the real power.

The only Linux thing not on Arch that I still run is a NAS I installed years ago with Centos and that still runs fine (but these days I would NOT use Centos for anything new.)

Why not FreeNAS?

[RoGeorge]

once you can deal with Arch you can jump in any other distro with ease

Yes, for sure.  Arch seemed to me more pragmatic.  Gentoo is even more hardcore than Arch, and if you want the ultimate learning experience about Linux, try LFS (Linux From Scratch).

None of these will be as clear and as straightforward as FreeBSD, IMO.

Though in practice I'm using Kubuntu on ZFS root, and FreeBSD only as a second OS.  Some things I need are not yet ported in FreeBSD, and/or are very hard to setup, but by the trend of Ubuntu and systemd, and now super-systemd (or something, forgot the naming, which will give one even less control to their own PC) won't be long until Linux as we know it will go extinguished.  Then, FreeBSD will probably become for Linux what Linux is now for Windows.


Later edit:
--------------
Found the name, systemd-supremo or something 
https://www.theregister.com/2022/10/26/tightening_linux_boot_process_microsoft_poettering/

[Nominal Animal]

I'd go so far as saying any experience with using multiple different OSes in anger helps.
They're complicated and different enough that just memorizing where things are will not really work, so one has to learn how to use them more effectively, with less cognitive/memory pressure, and focus on the task at hand: instead of looking for "where is that tool I like to use to do this", you start looking at "okay, how can I easily do this here right now".

I used to be quite active at LinuxFromScratch in its earlier days.  It is an interesting look at how to rebuild an entire Linux distribution from scratch.  It is useful learning experience for those who want to learn how to put together custom Linux setups on embedded devices, even if one ends up using something else as a basis.  Looking at the FreeBSD innards, from the sources and build system and init system upwards, should give a similar understanding; I just personally went with the LFS approach myself.

Which brings me to Debian, the core under many a Linux distributions, and its continuous fork Devuan, which wants to keep closer to the Unix philosophy, and keep more strictly to user choice, by supporting several init systems, excluding only systemd due to its rather aggressive assimilation of other subsystems into itself and making it a single point of failure (or security risk) for entire systems.  A minimal Debian/Devuan system can be a good basis for embedded devices and appliances, but you do need to know more about the bits underneath, so experience with LFS helps see the details when things don't actually work when at the package level they look like they should.

There are also dedicated Linux distributions for network appliances like routers, chiefly OpenWRT.  It has its own build and package systems, that do take a while to get a firm grip on.

Finally, there are experimental things like kFreeBSD-GNU, which is what you get if you take the FreeBSD kernel, and bolt on GNU userspace based on Debian packages on top.  I'm sure many are tempted to call it an abomination, but to me, it is an experimental thing done mostly to help understand the practical effects due to the differences in the kernels and userspaces.  You could say Microsoft WSL falls into the same category (although it is closer to counter-Wine); useful for many things, but definitely imperfect.

[SiliconWizard]

I've been using Arch as my sole distro for years...

Call me crazy, but I read the entire FreeBSD manual and it was so nice. Definitely an eyes opening experience, after years of DOS and Windows- I have no fear of FreeBSD anymore.
I even built my custom kernel.... That said I am not an expert, FreeBSD is so immense.... and I have limited time to test things out.

I am trying to repeat the same with Linux, Arch seems the best technical documented out there. I will setup a virtual machine and slap Arch on it, after I read through the Arch wiki....

Correct me if I am wrong, but once you can deal with Arch you can jump in any other distro with ease.

Well, a number of distros are more barebones and more complex to use than Arch.

Knowledge is the real power.

Sure. Well, the power is not in knowledge itself but in the ability to gather knowledge .

The only Linux thing not on Arch that I still run is a NAS I installed years ago with Centos and that still runs fine (but these days I would NOT use Centos for anything new.)

Why not FreeNAS?

You could have replaced that with 'Why not *?'
Never sure how to reply to this kind of questions.

[Zucca]

Many thanks, I am enjoying the discussion with a glass of bourbon in my hand.

LFS looks like a extreme painful way to learn.... I keep it in my mind if I want to go deeper... Arch seems to me the right sweet middle spot.

[Zucca]

You could have replaced that with 'Why not *?'
Never sure how to reply to this kind of questions.

Probably, why touch it if it works for me and I have 1000 other things to do?

[spanakop]

Why not try GhostBSD in a VM. You can test the hardware support first before installing. GhostBSD is basically FreeBSD with Mate automatically installed and configured straight from the installer. It has a graphical installer, and several apps to help you manage your updates and system. Of course you can go via CLI if you want. https://www.ghostbsd.org/

It has its own ports tree that is mostly a copy of FreeBSD's ports. You can think of GhostBSD as a kind of rolling release type of system.

This guy has a very good website on FreeBSD, which is still applicable to GhostBSD - https://vermaden.wordpress.com/

As others have mentioned, FreeBSD has a very good handbook, it does has Linux and Wine support if you want it, pretty good hardware support, although not as good as Linux and I find the community less toxic than Linux. The filesystem is simpler to understand and if it matters to you, does not have SystemD anywhere.

I have been using GhostBSD on my Lenovo T470 for a while and love it after being an openSuse user for many many years.

OpenBSD is more security focused, I believe does not have as good hardware support as FreeBSD and not as fast.
NetBSD is focused on more of a wider range of hardware than either of the above, but does not have as much software available (maybe wrong here)

As mentioned earlier, the BSD's are released as an entire OS i.e. kernel and base system apps. Where as Linux is really just the kernel. Your distro of choice then has to package and patch everything else together.