Announcement

Collapse
No announcement yet.

Ubuntu Blog Talks Up Rust Schedulers, Potential For Micro-Kernel Design Future

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

  • #41
    Originally posted by pabloski View Post

    Running a nuclear power plant system a complex enough scenario? https://www.itbusiness.ca/news/nucle...l-time-os/9084

    do you want to add air traffic control systems in the mix?
    Where is either of these scenarios even close to being complex? Especially the control of a nuclear power plant is very simple. Accidents never happen because the control was too complex, only because of human failure. Be it ignoring obvious warning signs or lack of maintenance. And air traffic control is simple enough that it can easily be done by hand if the need arises. Sure, it's much easier to have all information available on a screen and have that information be updated in realtime, but that's about it. So unless you want either job to be done by absolute novices, it's far from complex, especially for a computer. The only requirement you have is reliability, because errors can have disastrous outcomes, risking actual lives.

    Also these systems are the definition of specially tailored niche systems, a point you probably removed from the quote "by accident".

    On the other hand, the Linux Kernel as a (more or less) monolithic kernel has to be able to run on any hardware you put it on and run any software you throw at it. Both are degrees of complexity your examples will never have to fulfill.

    Comment


    • #42
      Originally posted by piotrj3 View Post

      Another user who thinks that you can use every std feature in kernel. (no you cannot). Everything with exceptions under the hood is not usable for example. Kernel memory management is a bit different as well. When C++ improves memory safety, there is also runtime cost to a lot of C++ memory safety features. Meanwhile Rust memory safety except array going out of bounds is entirely compiler based.

      Chrome or Firefox code base in C++ etc. are all the time updated codebases with fuzzers around and so on, and google technically by their hiring criteria takes very good engineers (or at least used to). Yet a ton of CVEs are allocated to those memory safety issues.
      non sense, obviously going from C to C++ you wont have exceptions at the beginning, but kernels with exception handling do exist. Not having all STL features implemented at the beginning is a transitioning strategy not a long term incapability. And chromium is a piece of trash coded by commies that dont know what they are doing.

      Comment


      • #43
        Originally posted by pabloski View Post

        Running a nuclear power plant system a complex enough scenario? https://www.itbusiness.ca/news/nucle...l-time-os/9084

        do you want to add air traffic control systems in the mix?
        Nope, it's not even comparable to database workloads where qnx would just scream in pain. When Linux gets hard real time capabilities you'll see switch on the market. Should I mention NY Stock Exchange?

        Comment


        • #44
          Originally posted by pabloski View Post

          the point is that microkernels are smaller, so less complexity means less bugs and better robustness

          this is why the kernel is part of the TCB
          When microkernel is smaller it just means it supports much less on its own. Has less drivers, file systems and other features. It's nothing, but small toy aimed for specific workload. Put it into serious workload on big SMP system and everyone will see how laughable performance it has.

          nuclear power plant doesn't count much if it only tracks some simple data.
          Last edited by Volta; 28 February 2024, 03:22 AM.

          Comment


          • #45
            That Tanenbaum–Torvalds debate is so old now as Linux now has both modular and microkernel abilities, showing how cheap is talk compared to work done.

            Comment


            • #46
              Originally posted by piotrj3 View Post

              Another user who thinks that you can use every std feature in kernel. (no you cannot).
              First, I was pointing out a general disinformation about the C++ language on pabloski's part. I never said anything about using C++ in the kernel. Congrats to you for throwing in distractions.

              Second, even if you want to argue against using C++ in the kernel, "not having std there" is the weakest argument of all. You don't have a C++ library there because C++ is not used and not the other way around. When Linus started writing the Linux kernel, nobody said "oh but you cannot do that because you can't use many of the C standard functions there". Now the kernel has many support libraries for C users. Same thing with Rust: Rust could only be integrated into the kernel by modifying standard Rust behavior and its std lib. If C++ ever came into the kernel, the kernel would of course receive classes for safe pointer management, data structures, and a lot more. And yes, to avoid exceptions, and to tailor it more to kernel-needs in other aspects too, it wouldn't be 1:1 the same as the C++ standard library. So what? C received tailored support libraries in the kernel, Rust received tailored support libraries in the kernel, but for some magical artificially contrived reason, C++ couldn't? This is a dumb argument (not to mention this wasn't even what we were talking about).
              Last edited by ultimA; 28 February 2024, 07:21 AM. Reason: Typo.

              Comment


              • #47
                I am going to post this in the threads that relate to the articles about Rust use because it is relevant,

                The White House is urging developers to stop using C and C++ and instead use Rust:

                Biden administration calls for developers to embrace memory-safe programing languages and move away from those that cause buffer overflows and other memory access vulnerabilities.

                Comment


                • #48
                  Originally posted by sophisticles View Post
                  I am going to post this in the threads that relate to the articles about Rust use because it is relevant,

                  The White House is urging developers to stop using C and C++ and instead use Rust:

                  https://www.infoworld.com/article/37...p-c-and-c.html
                  They must think very highly of their developers. Quite the opposite of your usual US arrogance, that their people can't do any wrong.

                  Of course this is utter BS, Rust is for people lacking proper knowledge. But if you actually know your C/C++ there's no real reason not to just keep using what you know best. Of course besides just buzzwording.

                  Comment


                  • #49
                    Originally posted by ultimA View Post

                    [...] Smart pointers in general are not the only way to achieve memory safety [...]
                    Blazingly 🔥 fast 🚀 memory vulnerabilities, written in 100% safe Rust. 🦀
                    Blazingly 🔥 fast 🚀 memory vulnerabilities, written in 100% safe Rust. 🦀 - Speykious/cve-rs

                    ​

                    Comment


                    • #50
                      Originally posted by Artim View Post

                      Rust is for people lacking proper knowledge. But if you actually know your C/C++ there's no real reason not to just keep using what you know best. Of course besides just buzzwording.
                      Since there are so many memory vulnerabilities reported all the time it follows that most programmers don't actually know C/C++, so it is useful for most programmers to use Rust. Interesting.

                      Comment

                      Working...
                      X