Slashdot Mirror
UbuntuBSD Is Looking To Become An Official Ubuntu Flavor (softpedia.com)
prisoninmate quotes a report from Softpedia: UbuntuBSD maintainer and lead developer Jon Boden is now looking for a way for his operating system to contribute to the Ubuntu community and, eventually, become an official Ubuntu flavor. Just two weeks ago, [Softpedia] introduced the ubuntuBSD project, whose main design goal is to bring users an operating system powered by the FreeBSD kernel while offering them the familiarity of the Ubuntu Linux OS. Right now, ubuntuBSD is in heavy development, with a fourth Beta build out the door, and it looks like the developer already seeks official status and wants to contribute all of his work to the main Ubuntu channels. [Canonical has yet to respond.]
I got into a debate with my former Linux users group on this when a fork of Debian hit a half decade ago with FreeBSD.
Everything from gnome to pulse audio to SystemD is integrated in Linux. People act as if you can swap the kernel out and still run or even compile anything. I am shocked anything works at all with gnome on non Linux platforms as things are so proprietary and tight. Yes it's gnu, but what I mean by proprietary is Linux and not Unix standard way it does things since 2006
FreeBSD is a server oriented OS as far as I am concerned
http://saveie6.com/
Tastes like SHIT!
The sad part is you apparently know what shit tastes like.
But to the topic at hand, FreeBSD, which is more of a server OS, is going to have to make the changes to "become" Ubuntu as opposed to Ubuntu becoming BSD. Most who install Ubuntu expect to have their devices ans peripherals running after install.
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
Because the kernel is what is wrong with Ubuntu. Or Linux in general.
This sounds a whole lot like the Debian GNU/kFreeBSD distro! It was the Debian userland running on top of the FreeBSD kernel.
For those who don't remember, it was yet another casualty of systemd being forced into Debian, which caused GNU/kFreeBSD to be effectively killed.
Systemd is the worst thing that has happened to Debian. Not only did it cause great pain and upheaval for many long-time Debian users, especially those who need reliable and stable systems, but it also killed projects like Debian GNU/kFreeBSD, which had little to do with Linux to begin with.
It's kind of funny, in a sad way. The very action (using systemd) that killed the integration between the Debian userland and the FreeBSD kernel ended up rendering the Linux-based Debian distro unusable for many, forcing them to switch directly to FreeBSD itself!
Systemd is the worst thing to have happened to Linux, and the best thing to have happened to the BSDs!
And whipslash, if you're reading this, can you please get rid of the goddamn posting limits? I've posted maybe once or twice in the last 12 hours, if not longer, yet I kept get the fucking "You must wait a little bit before using this resource; please try again later." error message. Ditch the idiotic posting limits, please.
Haven't ports of debian already been tried to alt OSes such as FreeBSD, HURD and OpenSolaris with a tiny number of users to support and develop the platform?
Please stop trying to taint *BSD with the nasty ass disorganized chaos of GNU.
*BSD are nice, organized, predictable, maintainable OSes with good clean utilities that work in reliable and predictable ways.
GNU is a piss pot of everything doing everything THEY want and not giving one flying fuck about how anyone else might like it:
Examples:
Not using /usr/local ... except different so that nothing actually will work out of the box across them without tweaks
systemd
4 trillion flavors of GNU/Linux that all work exactly the same
The Linux desktop
Stop trying to make a fucked up BSD, its not a god damn shitty GPL project, its actually open source.
JUST TAKE THE KERNEL PARTS YOU WANT AND LEAVE *BSD ALONE. WE DON'T WANT YOUR CORRUPTION.
And no, I'm not joking.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
How should kernel developers go about debugging a crash caused by a proprietary kernel module?
Please get it through your head wannabe Linux distro-devs , BSD is not Linux! Please stop attempting to fracture the BSD community with hundreds of distros!
If you want a easy desktop system, put your development efforts into improving PCBSD not UbuntuBSD.
If you want Arch pacman or Gentoo ports (don't know the name), it's mostly already covered by FreeBSD pkg and ports respectively. If you don't like them, just write new package managers for FreeBSD, there is no need for ArchBSD or GentooBSD.
If you want an outdated GPL userland, just install outdated GPL userland stuff on your system, there is no need for DebianBSD.
In the past the BSD community generally only split when it had good reason. If you're not rolling your own kernel or there isn't a commercial implementation of your new "distro" then you probably shouldn't do it. Go play with Linux.
What are these "good reasons?"
FreeBSD forked off of 386BSD when it died.
NetBSD forked off of 386BSD focused on multi-architecture support.
DragonFlyBSD forked off FreeBSD due to threading implementation.
OpenBSD forked off NetBSD to focus on heavy security.
These "Big 4" are completely different OS's, they all have their own kernel, they all have their own userland. They do share a lot, but they're not the same OS's.
Then you have Open Source Commercial OS. All of these are based off FreeBSD.
PC-BSD - iXsystems desktop OS.
FreeNAS - iXsystems NAS OS.
PFSense - Electric Sheep Fencing firewall OS.
And there are probably more that would go into that category.
If you don't fit into either slot. (New OS or Commerical) you probably should contribute to existing platform than searching for your pointless personal glory of being a distro maintainer.
> Yeah we could all run Windows 2008 Server and call it a desktop,
Windows IS a desktop OS, not a network OS. Notice it's not usable except by clicking desktop icons? You can certainly argue that it's a poor desktop because not many applications are been installed by default , but the operating system is desktop through and through. A server OS doesn't require rebooting every week or every month. Average uptime for my servers is probably about three years, because I physically moved them a few years ago. A server OS can handle hotswap hardware. I swap drives regularly, and we've even hotswapped a CPU. You don't reboot a server three times when > Yeah we could all run Windows 2008 Server and call it a desktop,
Windows IS a desktop OS, not a network OS. Notice it's not usable except by clicking desktop icons? You can certainly argue that it's a poor desktop because not many applications are been installed by default , but the operating system is desktop through and through.
You don't reboot a server every week or every month. A server OS can support server hardware, which includes hotswap. You damn sure don't reboot a server just because you swap out the backup drive - hotswap SAS and SATA are a bare minimum for a server OS. We've even hotswapped a CPU on a server. The server OS (Linux) didn't skip a beat, because servers support hotswap. That's hotswap PCIe, not sometimes USB works if you properly eject it first.
A server OS doesn't implement "security" by HIDING files in the GUI. A server OS has mandatory access control.
I could go on and list a dozen more things, but the bottom line is that Windows is a very successful desktop operating system. One originally developed as a user-friendly shell for Disk Operating System, so named to distinguish it from all of the others at the time, which were network operating systems.
I don't know what happened there. It appeared that my browser deletee what I had typed, so I typed it again. Somehow it ended up posting both copies. I'll blame it on many mobile OS.
Frills like audio, usb support, working wifi or ethernet cables, multiple had drive setup or audio card support are for newbs. Real computer users don't bother with any of that.
not giving one flying fuck about how anyone else might like it
Apple products are all about industry standards. Dell and HP support proprietary dock connectors on their systems while Apple uses industry standard thunderbolt for expansion.
FreeBSD, which is more of a server OS
FreeBSD officially supported platforms include XBox and embedded ARM systems.
you really could not be more wrong even if you actually tried
as much as Solaris is BSD
FUCKING IDIOT
SunOS was BSD based, they RIPPED IT OUT, and dropped in System V when they re-branded as Solaris
Windows has more BSD code in it than Solaris
Uninstalling a buggy driver is not debugging. Which is what you are reduced to when you do not have the source code.
yeah, there's no such thing as "customer support"
no chance in hell.
It's just Debian/Ubuntu with a FreeBSD kernel, no other signs of BSD at all. Rather disappointing.
FreeBSD, which is more of a server OS
Servers like Sony's PS3 (Vita OS) and PS4 (Orbis OS), both of which are based on FreeBSD?
Most who install Ubuntu expect to have their devices ans [sic] peripherals running after install.
Citation Needed!
You know what? Linux isn't all rainbows and unicorns. A recent update to the Fedora kernel broke my dual monitor setup. Yes, the kernel. Reverting to an earlier kernel with no other changes restored my dual monitors. Gnome 3 Desktop has routine breakage. Yeah, don't tell me that Fedora isn't Ubuntu, I already know that.
Part of the difference might be that there are actual companies selling Linux. Companies with lawyers, who can approve NDAs, that allow kernel developers to get early access to new devices, so Linux tends to get support for new stuff earlier. *BSD's kernel developers may not always have that kind of luxury.
I've used FreeBSD since the beginning, and 386BSD before that. I've never had bleeding edge hardware and I've never had a problem with FreeBSD supporting my hardware.
You test the binary blob against the requirements of the ABI. If the kernel crashes harden and protect the interface further.
At some point, you end up having to build all the overhead of a microkernel in order to "harden and protect the interface" against popular yet defective binary drivers that violate "the requirements of the ABI".
FreeBSD, which is more of a server OS
Servers like Sony's PS3 (Vita OS) and PS4 (Orbis OS), both of which are based on FreeBSD?
Most who install Ubuntu expect to have their devices ans [sic] peripherals running after install.
Citation Needed!
SO you ayr sAyn' that peepule d'unt expect they're nstuls to work? It's to mech! Gow fguure/
The shepherds did so well protecting the flock that the sheep no longer believed that wolves existed.
I actually read many (not all) of the comments on this topic, and - unusual even for Slashdot - there was NOTHING of value. This topic will appear at Hacker News, and it will contain at least 50% intelligent and thoughtful remarks.
Dropping in on a technical Slashdot thread these days is like visiting a religious convention of snot-throwers being hosted by Youtube. As a result, I (and I'm sure many others) read very little of what all of you zealots have to say. If that's your desire, congratulations. Enjoy your porn pit.
"You must try to forget all you have learned. You must begin to dream." -- Sherwood Anderson
That is a directory for things that local admins wish to install. Why do you pollute it with vendor stuff?
Putting stuff there is almost as bad as /tmp or /root or /usr/ken or /usr/dmr would be. It's not your directory. Don't touch it.
It's funny to me that some people delight in pointing out that Linux isn't a wonderful desktop personal computer operating system, but then get all uncomfortable with the idea that Windows IS a desktop PC OS . Personal computers revolutionized the world, and PC desktop operating systems are hugely important.
> Define Network OS
It's part definition and part heritage. I'll try to cover both, but first I'll go for a short-form, example. Unixes won't boot without their network stack. The graphical windows on the screen are a bolt-on afterthought, delivered through a SOCKET. If you want to configure printers on a Unix, you access 127.0.0.1:931 - you constantly use the network to access the local machine.
Definition wise, before defining network operating system, we should make one point about the definition of operating system, and operating system families generally. The family to which an operating system belongs isn't determined by one or two userland applications. Installing a graphical web browser on an instance of a network OS doesn't turn it into a personal computer desktop OS. (In fact, a browser is a graphical tool for accessing the NETWORK, being a network client using network applications). Likewise, deleting the browser, or the desktop calculator, from a desktop (PC) OS doesn't make the OS stop being a desktop OS. Windows is an operating system. Windows 7 Home and Windows 7 Ultimate are two PRICES, mostly for userland applications, not two operating systems.
Network are designed with network access being a basic assumption. They don't have network DLLs added in some version, they start with the assumption that most users will be accessing the system over the network. The local console may be extremely limited, because they the local keyboard and monitor isn't the standard way of using them. That's because the whole point of developing network operating systems, starting with MULTICS in 1964, was that they are multi-user, with several concurrent users using them over the network. Often, primary storage and other important components are accessed over the network.
On the other hand, PC (personal computer) operating systems generally assume a local user is the typical use case. A network isn't required or assumed, local storage is the norm.
On MULTICS it was also common to remove half of the components each night and assemble them into a second system. At night, half of the CPUs, half the storage, half the RAM, etc would be removed, a devel/test system assembled, then in the morning all of the parts would be combined again into one large computer. The computer wasn't shut down during this process - programs continued to run as CPUs were added and removed. That's one defining characteristic of an OS family related to NOS - the server operating system.
From MULTICS came UNIX, and from UNIX sprang BSD and Linux. (Linux of course re-implemented Unix, rather than copying source code files). This OS family has been based on multiple concurrent users, over the network since it began in 1965, 50 years ago. Of course with multiple concurrent users comes the need for securing one user's file and processes from access or interference from another user, so that was always built in to this OS family. Computers cost hundreds of thousands of dollars each, so multiple concurrent users was a bit important.
Teen years later, electronics had advanced enough that for just a few thousand dollars you could build an electronic word processor with 4KB of RAM and a 1 Mhz CPU. Obviously UNIX wouldn't run, with user software, in 4KB of RAM. Hell just the networking code alone was 4KB. So a couple of really smart people had a great idea. They could build a PERSONAL computer, for one user at a time, by putting together a tiny OS that would run on word processor hardware. It wouldn't need anything multiuser - no networking, no security, etc. With just one home user at a time, it didn't need enterprise grade reliability. It DID need to fit into 4KB of RAM, and boot a from a 140KB st
Paragraph four, which basically defines network operating systems (as I'm using the term), was murdered by a few typos. It should read:
Network operating systems are designed with network access being a basic assumption. They don't have network software added on in some version. They start with the assumption that most users will be accessing the system over the network. The local console may be extremely limited, because most users won't be using the local keyboard and monitor. That's because the whole point of developing network operating systems, starting with MULTICS in 1964, was that they are multi-user, with several concurrent users using them over the network. Often, primary storage and other important components are accessed over the network. With the CPU in one place, the disk somewhere else on the network, and the keyboard somewhere else, the system can't really even come into being unless the network is up and working. Network operating systems are designed for that type of system as their natural environment. Another example is that I routinely install Linux on machines thousands of miles away, without using a KVM. Networked access is the norm, of course you can easily install the OS over the network.
It's funny to me that some people delight in pointing out that Linux isn't a wonderful desktop personal computer operating system, but then get all uncomfortable with the idea that Windows IS a desktop PC OS .
Define "wonderful desktop personal computer operating system", and then tell me the scope of what you're talking about. Right now the discussion started at the kernel. The Linux kernel is indeed a wonderful desktop personal computer OS. It's smooth, fast, polished, has features that if tweaked correctly provide a fantastic desktop user oriented connection to the underlying hardware. What the entire package that people call "Linux" is missing is a Desktop environment with polish that is free of bugs and frustration.
The same argument is used on the server side. Define what makes a server OS, and do it in a sensible way, not just everything needs to be built on a network stack. To make that claim you first need to show why the entire OS built on a network stack improves the environment for a server. Accessing a local printer via a network socket? Who gives a shit. Instead why not compare the CUPS performance for locally attached networks when exported over a network.
People like to cherry pick their arguments and definitions in these cases to make their own favourite the winner. The only thing that matters on the desktop is the experience to the user, the only thing that matters on the server is the the ability to perform functions related to being a server. At the kernel level Linux and Windows offer two different products highly tuned to each function, BSD provides one tuned to the server, unless you cound OSX then a BSD based kernel can be said to do both as well. At a higher level GNU/Linux does not have a very polished interface with easy to use software for the desktop but has a great list of features and software support for servers. At that same level Windows is good at both, regardless of the fact that it got to that point by starting out as a desktop OS.
>>> Define "network operating system"
[six paragraphs explaining network operating systems to you.]
> To make that claim you first need to show why the entire OS built on a network stack improves the environment for a server.
> Accessing a local printer via a network socket? Who gives a shit.
> Instead why not compare the CUPS performance for locally attached networks when exported over a network.
Because we're not talking "performance operating system", we're defining "network operating system". A network operating system is one that's based on the network, as a precondition. When the storage is at 10.0.2.3, the CPU is 10.0.2.6, and the console is 76.212.34.11, that's probably a network system. It's probably a good match for a network operating system.
When the console is connected via USB and DVI, the drive via SATA, and it's sitting on a desk, that's probably a desktop system. A desktop operating system makes sense.
When the 4" console is glued on top of the CPU and the storage is a soldered-on 2GB chip, that's probably a mobile system. A mobile operating system would be a good match.
Let's start with your first two sentences:
> you are naive if you think that you can avoid rebooting for all Linux updates. Obviously a kernel update is going to require a reboot
You are mistaken. What is "obvious" to you is not in fact true. Since the beginning, we've been able to upgrade the majority of the kernel with nothing more than "rmmod foo && insmod foo". I've been doing that for twenty years. Seventeen years ago, we gained the ability to upgrade the kernel core live. That's now a default feature of kernels from Redhat and other major distros.
I could go on and similarly address the rest of your misinformed post, but I think it's sufficiently clear that you're at least thoroughly misinformed, if not actually suffering from Stockholm syndrome, defending the organization which holds your business captive.