Announcement

Collapse
No announcement yet.

Haiku OS Powered By BSD? It's A Possibility

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Haiku OS Powered By BSD? It's A Possibility

    Phoronix: Haiku OS Powered By BSD? It's A Possibility

    François Revol presented at FOSDEM this weekend about the prospects of Haiku OS ever becoming a BSD distribution. Haiku OS, the well known BeOS re-implementation, does currently rely upon some BSD components but more integration is possible...

    http://www.phoronix.com/scan.php?pag...SD-Possibility

  • #2
    I don't get it... Why threaten to adapt an ancient unusable kernel for an OS that it isn't suitable for? The BSD's are all dying because of their own arrogance. They think that if they do all the work themselves that they can benefit the world. But all that really happened is commercial entities abused them and now they don't have the resources to survive.

    Adopt a copyleft. It's the only viable option for survival. Not the BSDs.

    Comment


    • #3
      Originally posted by duby229 View Post

      Adopt a copyleft. It's the only viable option for survival. Not the BSDs.
      Switching to a Linux kernel would be a great solution. The Haiku OS kernel is nothing special, whereas the original BeOS kernel was amazing.

      Comment


      • #4
        Originally posted by duby229 View Post
        I don't get it... Why threaten to adapt an ancient unusable kernel for an OS that it isn't suitable for? The BSD's are all dying because of their own arrogance. They think that if they do all the work themselves that they can benefit the world. But all that really happened is commercial entities abused them and now they don't have the resources to survive.

        Adopt a copyleft. It's the only viable option for survival. Not the BSDs.
        Ummmm.... Are you fucking delusional?

        FreeBSD and OpenBSD, at least, receive millions in donations combined each year, from big companies like Microsoft, WhatsApp and NetFlix to continue developing.

        There's nothing ancient or unusable about a BSD kernel, compared to the Linux kernel. The Linux kernel does have more features, but thats irrelevant as half of them are still broken and unusable, almost everything promised in the current BSDs works as intended. There's a difference of fit and finish.

        You can not like BSD, that's fine, but if you're going to trash it, trash it for things that make sense, such as slower and more conservative development.

        Linux is still not a complete ecosystem, unlike each and every BSD which is in and of itself, independent. In almost every Linux distro, there's no separation of the base system from what is installed post install - yum or aptitude can and will uninstall a kernel and leave you with an unbootable system. Pkgng won't, freebsd-update won't, and make buildworld certainly won't.

        I use both, I use BSD more, but I have Linux on my IA-64 gear - it does its limited purpose very well. But BSD works for us.

        Anyways, copyleft is irrelevant here, Haiku has problems because its a monumental task to recreate BeOS - just as ReactOS has problems with Windows.

        Adopting the Linux kernel wouldn't help due to licence incompatibility, and I should remind everyone here that the GPL has produced its own fair share of monolithic, unmaintainable static projects: Screen, for example, went without development for nearly a decade due to its code becoming unmaintainable. The "solution" was to wrap Byobu around it, which isn't really a solution at all. Its like painting over stress fractures on a piece of metal - it doesn't fix the underlying brokenness and that is what I have a problem with.

        Comment


        • #5
          I don't think Haiku should adopt a Linux kernel. That's not what I said. It's pretty obvious you're a nutjob though. Just in your one post I've read a number of things that make me question your competency or at the very minimum your sanity.

          EDIT: Why in the hell would any system admin deploy an IA64 system? They cost thousands of dollars more than anything else available at that time and performed terribly across the board. It -is- the single most incompetent thing an admin could deploy at that time. No doubts and no questions. If an IA64 platform were given to me I might find some use for it, but reality is that I would much better performance on almost anything else.
          Last edited by duby229; 01 February 2016, 02:26 PM.

          Comment


          • #6
            I don't think he was suggesting that Haiku/kFreeBSD was all that likely but just that it might be possible... and that reusing more drivers and components from BSD *is* likely... mainly because the device driver interface on BSD is fairly stable ... in effect the Haiku C++ kernel (newos) would become a BSD compatible kernel for the sake of drivers and portability to-from BSDs.

            Comment


            • #7
              Is this the 5-minute argument, or the full half-hour?

              I think Haiku should keep its NewOS-based kernel. There's no reason it can't stick with that while making use of BSD code for drivers. I can understand though, if they would save time in the bigger scheme of things by doing so (switching). But then there would have to be someone ready to take on board the up-front time cost of doing it.

              I used to think that the Haiku kernel was not quite as performant as some people made it out to be - until I found out that youtube videos run smoothly under Haiku on my EEE901 (whereas they judder and freeze under OpenBSD and FreeBSD). That won't be entirely down to the kernel, but Haiku's kernel is no slouch, either.

              EDIT: Having read the slides now (heh).. I don't see the kernel changing any time soon, at least not yet.
              Last edited by philcostin; 01 February 2016, 02:56 PM.

              Comment


              • #8
                Originally posted by duby229 View Post
                I don't think Haiku should adopt a Linux kernel. That's not what I said. It's pretty obvious you're a nutjob though. Just in your one post I've read a number of things that make me question your competency or at the very minimum your sanity.

                EDIT: Why in the hell would any system admin deploy an IA64 system? They cost thousands of dollars more than anything else available at that time and performed terribly across the board. It -is- the single most incompetent thing an admin could deploy at that time. No doubts and no questions. If an IA64 platform were given to me I might find some use for it, but reality is that I would much better performance on almost anything else.
                I'm perfectly sane - I have no idea what you're on about. I apologize as I was directing the Linux kernel quip to the poster below you. Calm down, I jumped a little because I found your remark about BSD to be overly inflammatory, and I'm totally guilty of that.

                And I deploy IA-64 systems because I get paid to maintain customers with no direct upgrade path - their applications can't run on x86 or x64 and porting them is more money than they're willing to fork over right now, their software takes advantage of several features of the VLIW-like architecture of IA-64, RHEL 5 is still supported and usable on them, and currently there's no x64 equivalent to the NUMALink system that the Altix utilizes. They'll eventually migrate to POWER, it looks like and use Linux on POWER, because POWER's capabilities are more suited to their use. By the way, when these servers were first purchased in early 2000's there was no x64, no ccNUMA at all, Alpha was already down the tubes, POWER wasn't competitive and MIPS, which they migrated from, was a dead end. I was brought on in 2015 to keep maintaining their infrastructure. They of course have x64 servers in their fleet, but they're an even split between FreeBSD/ZFS storage servers, and Ubuntu Server ( pre-systemd ) for webfacing applications. The 128 IA-64 fleet is split into several NUMALinked blocks, with each block of servers acting as a single entity, and to be honest, they're really easy to maintain.

                And cb88, I do believe making some of the driver interfaces more BSD-like would be beneficial for ease of porting from the BSD kernel - I do personally believe that they're better off sticking with their current kernel design. Going the route DFBSD has with making some interfaces more similar to another kernel for the sake of porting is smart, because it doesn't really affect anything in the long run.

                Comment


                • #9
                  Originally posted by TeamBlackFox View Post
                  FreeBSD and OpenBSD, at least, receive millions in donations combined each year, from big companies like Microsoft, WhatsApp and NetFlix to continue developing.
                  If that's the case, then how this happened?
                  http://www.undeadly.org/cgi?action=a...20140114072427

                  Comment


                  • #10
                    Originally posted by TeamBlackFox View Post
                    Linux is still not a complete ecosystem, unlike each and every BSD which is in and of itself, independent.
                    T'was never meant to be an ecosystem in and of itself. It's a project that other entities (should they be corps or NPOs or your uncle's cousin's brother's neighbor's college roommate) take and make their own ecosystem (or projects) around. It's an entirely different methodology and world from the BSDs. BSDland is a bunch of neighboring traditionalistic monarchies, with some differences. Linuxland is a libertarian stateless entity (or anarcho-communism might fit better, but I digress). Of course there'd be a schism in world views and people from each camp jabbering on and on about how their camp is right and the other is much more flawed. Much like you're doing right now.

                    In almost every Linux distro, there's no separation of the base system from what is installed post install - yum or aptitude can and will uninstall a kernel and leave you with an unbootable system. Pkgng won't, freebsd-update won't, and make buildworld certainly won't.
                    Too bad "almost every Linux distro" doesn't use yum (outside of RPMland) or Aptitude (outside of Debianland). You're making a sweeping generalization about Linux by using 2 package managers. Ridiculous.

                    Pacman won't, unless you tell it to uninstall the kernel (should you install only 1) while not touching its dependencies, thus purposefully making an unbootable system.
                    Portage won't, unless you either do exactly what I described previously or muck up your ebuild or kill your tree. Any case being your own damn fault.
                    Nix refuses, since it always makes sure your OS has a clean state to boot from.
                    I've never heard of Zypper "inadvertently" uninstalling the kernel, but btrfs snapshots on /, plus incremental backup, makes that not an issue, even if it is true.
                    I don't know if it's even possible for apt-get to David Copperfield a kernel, unless you have a PPA for, say, the latest kernel+drivers. But that only affects the old kernels after a certain number of kernels are on the system (not allowing more than 4 versions, for example).

                    Are you saying that, because of lack of separation of base system and explicitly-installed packages, that any Linux distro is capable of killing itself? If that's what you're saying, then that's simple reasoning...because it's simply wrong. It would take monumental user error or upstream breaking their distro for that to happen. The former can happen if the user is an idiot, and, while the latter has happened before, it certainly didn't happen because l'il baby Linux didn't listen to big daddy BSD.

                    I use both, I use BSD more...
                    Lol trust, it's not a feat of mind to glean that much from you.

                    Haiku has problems because its a monumental task to recreate BeOS - just as ReactOS has problems with Windows.
                    Haiku has problems because they're doing OS development with an app-development-sized team. Plus, their methods are both bass-ackwards and highly inefficient.
                    ReactOS has problems because they're fucking mental. They just don't know how to manage a project of any scope.

                    Adopting the Linux kernel wouldn't help due to licence incompatibility, and I should remind everyone here that the GPL has produced its own fair share of monolithic, unmaintainable static projects: Screen, for example, went without development for nearly a decade due to its code becoming unmaintainable. The "solution" was to wrap Byobu around it, which isn't really a solution at all. Its like painting over stress fractures on a piece of metal - it doesn't fix the underlying brokenness and that is what I have a problem with.
                    On the other hand, never have I had such an easy time implementing services in Linux before I started using Arch. Thank GPL for systemd.

                    Comment

                    Working...
                    X