Announcement

Collapse
No announcement yet.

OpenBSD Sucks? Thoughts From One Of Their Developers

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

  • #11
    Anyone else having the site with the slides pretending it's the dial-up era all over again?

    Comment


    • #12
      Originally posted by dante View Post
      from the security point openbsd rocks like no other os but from other points maybe it not so good
      Openbsd, and every other monolithic kernel, are far behind the security of seL4.
      Not even close.

      Comment


      • #13
        Originally posted by liam View Post
        Openbsd, and every other monolithic kernel, are far behind the security of seL4.
        Not even close.
        Sadly it's going to be a while before microkernel OSes ever stand a chance of adoption in the OS sphere. They're just not developed enough to be there.

        Comment


        • #14
          Originally posted by Luke_Wolf View Post
          Sadly it's going to be a while before microkernel OSes ever stand a chance of adoption in the OS sphere. They're just not developed enough to be there.
          Blackbarry OS 10 is a microkernel os(QNX). It is pretty secure. QNX is used quite a bit in automation.

          Comment


          • #15
            Originally posted by Nobu View Post
            Is that why I always liked Comic Sans? o_o
            Quite possible!

            It's not quite as good as fonts designed for it (like opendyslexic), but the varied shapes and stroke weights apparently make it very readable.

            Comment


            • #16
              Originally posted by liam View Post
              Openbsd, and every other monolithic kernel, are far behind the security of seL4.
              Not even close.
              Unless someone builds a CPU with a zero-cost security transistion, micro-kernels will always be slower than monolithic kernels, and will therefore always be second-class. Performance is always important. On mobile CPUs every milliwatt counts and on desktop or server CPUs where performance is their whole reason to exist.

              Comment


              • #17
                Originally posted by Zan Lynx View Post
                Unless someone builds a CPU with a zero-cost security transistion, micro-kernels will always be slower than monolithic kernels, and will therefore always be second-class. Performance is always important. On mobile CPUs every milliwatt counts and on desktop or server CPUs where performance is their whole reason to exist.
                The newer l4 kernels have very fast context switches.
                Honestly, though, it doesn't matter. I was only commenting on monolithic kernel security, not performance.
                I maintain that for iot, something like sel4 is necessary. Arm is promoting their, closed source, ukernel mbed os, for just such devices. I'd much rather have the open source sel4 in that position.

                Comment


                • #18
                  Suxx for sure!

                  Of course OpenBSD suxx! Its nearly unusable for anyone but bunch of lunatics who do not care about doing anything useful at all.

                  These guys turned OS development and administration into religion where they pray their quasi-god Theo. And of course there are counterproductive rituals instead of well thought actions all the time. Its so damn productive to use CVS as of 2015. Even outdated SVN scrap is way too new for openbsd, not to mention things like GIT. That's what I call retarded and completely counterproductive project management.

                  They tell us:
                  missing of snapshot capabilities with the FFS file-system
                  Poor dullards! They are so retarded they even fail to understand 80s are gone as well as 90s gone too. It's not about snapshots, nuts! It is about overall filesystem design which got hopelessly stuck in early 90s disk/storage techs if we take a look on FFS. FFS is an ancient bones and dust of file systems world. Not anyhow comparable to modern designs, because people created countless improvements in storage techs and others used these. And of course snapshots are working best on recent filesystem designs, created from scratch, using copy-on-write as core part of design. Because when CoW is central feature of your FS, it creates snapshot out of pre-existing structures, completing snapshot operations in blink of eye. You can't just came, add couple of nuts and bolts to your MS-DOS aged shit and get decent snapshot support. You have to completely redesign it and then rewrite it. Something BSD nuts can never afford due to lack of resources. Especially in projects with totally crappy project management like OpenBSD.

                  Comment


                  • #19
                    Originally posted by liam View Post
                    I maintain that for iot, something like sel4 is necessary.
                    Listen my simple prophecy: Moore's law would go on, at least for some time. CPUs will be more powerful, RAM will have larger space, costs will go down and so on. At some point cheap and small systems will be able to run usual Linux and there will be very little point to make system less powerful than that. Nearly everyone will make IoT stuff out of common Linux devices, thanks to relatively simple development, good networking stack, plenty of drivers, numerous SoCs supported and so on.

                    In fact this time starts here and now: cheap, low powered, small, coin sized, SD-card-sized computers already have CPUs capable of running Linux and more or less supported, even by mainline kernel. Yet they have decent networking, up to being able to work as full-fledged wi-fi access point. Yet, it costs about $10 or maybe $15 at most.

                    I think it is pretty clear what will make IoT possible and useful: it will be Linux. Sure, maybe 0.005% of market will need some uber-secure, super-crash-resistant devices. But these are inherently costly, have awful R&D costs, lack features everyone expects these days and you can't expect high demand on such devices. Just like you can't expect to sell armored car to Average Joe: even if it can withstand bazooka shot, Average Joe hardly needs this feature and not willing to pay for that.

                    Good project management is all about keeping things adequately balanced. Something that BSD zealots and microkernel dreamers fail to do all the time.

                    Comment


                    • #20
                      Why would anyone install OpenBSD on something that has nVidia video card? And even if somebody does, why would they need a specific driver? It's a server OS and GUI is about the last thing it needs.

                      Comment

                      Working...
                      X