Announcement

Collapse
No announcement yet.

Linux Kernel's Preliminary Rust Code Seeing 64-bit POWER Support

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

  • #21
    Originally posted by Alexmitter View Post

    Then you just aren't good in C.
    Fefe wrote an excellent article about statements like this. He calls them "hybris of the programmer".
    https://heise.de/-4879795

    Comment


    • #22
      Originally posted by dragonn View Post
      Having two compilers for [rust] would be just a waste of dev resources.
      So clang is also just a waste of resource? Interesting viewpoint you have.

      Comment


      • #23
        Originally posted by Alexmitter View Post
        So was Go, the last hype language. And here we are now in the post Go world.
        ???

        Go is doing great. It's a language which makes perfect sense in certain domains, kernel development is not one of those and I haven't seen anyone argue that it is.

        As for Rust inclusion into the mainline kernel, I can't see it happen until the language has truly stabilized, I would also think that GCC Rust support would be a key factor. In other words, I think it will spend a very long time in Linux-next.

        Comment


        • #24
          Originally posted by all3f0r1 View Post

          BTW the "Rust fan-base" as you call it is actually very heterogeneous. You seem to imply that "Rust fans" are only composed of hype hunters (might be, to some degree), but as far as I can tell, it's mostly composed of a very broad range of programmers, coming from C, Python, ASM, JS...
          0. Fans are the guys that make the noise, without necessarily having written a single line of code. The guys that use Rust because they like it (for whatever reason) or because they can be more productive with it (also for whatever reason) are usually not fans but people with informed opinions. There is lots of point with discussing with the later but no point with discussing with the former.

          1. C and asm is programming, python and js is scripting. Both things are useful but different. There is also the thing I call "emulated programming", which is using scripting (usually python) in order to write something that looks like a slow program.

          Comment


          • #25
            Originally posted by mlau View Post
            the whole "it makes you code much more safe without having any performance downside" is hyperbole/marketing.
            No, it's not hyperbole nor marketing. Rust's memory ownership model is an improvement over *all* similar features in other system-level languages. It is significantly better.

            Comment


            • #26
              I still think that t[rust], is not, at least yet, there as a substitution for C..

              Comment


              • #27
                Originally posted by tuxd3v View Post
                I still think that t[rust], is not, at least yet, there as a substitution for C..
                I don't think that's the point though. Letting parts of the kernel being written in a (mostly) safe language is a good step forward, but I doubt any extensive rewrites of the kernel will be made any time soon for a variety of reasons (experience with C, Rust not supporting the obscure arch that Linux does, too many lines of code written, etc).

                The more realistic approach for long term improvement, would be to make the language safer with extensions or to use hardware assisted memory safety (such as memory tagging).
                Last edited by kvuj; 23 March 2021, 08:55 PM.

                Comment


                • #28
                  Originally posted by mlau View Post
                  the whole "it makes you code much more safe without having any performance downside" is hyperbole/marketing.
                  This is the biggest load of bullshit I have heard in a while. Rust's ownership model has been formally verified, of course there can still be bugs in LLVM or in other parts of the toolchain but by definition it will always be more safe than C.

                  There are also plenty of ports of C tools written into Rust and they

                  1. Are faster than the C equivalent while being more safe
                  2. Same speed as the C equivalent while being more safe
                  3. Slower than the C equivalent while being more safe however if the safety checks are disabled/removed (i.e. bounds checking on arrays) then the speeds are equivalent.

                  Comment


                  • #29
                    Originally posted by miquels View Post
                    I know, rust sectarians never read past the bold captions and loud slogans, so what exactly did you mean? Maybe this: Larry Wall famously promotes "the three great virtues of a programmer: laziness, impatience, and hubris".

                    Comment


                    • #30
                      Originally posted by reba View Post

                      So you're saying Reinhold Messner is a bad climber because he uses equipment to make his ascent more safe?
                      If you think that this is appropriate analogy, let me give you mine: suppose you are in need of surgical operation, who would you prefer to perform it? A seasoned surgeon who really knows what he is doing or a D-student, but with his hands tied in such way that he can not move sharp tool outside the operating area "ensuring" that his actions are "safe".

                      More of a silly analogy: the seasoned surgeon is operating in an established institution which was there for 50 years and will still be there in foreseeable future, the D-student is operating in a private commercial entity who only say "no warranties, but trust us, we know what we are doing" and "everything is subject to change in any time, but again, trust us, we know what we are doing".

                      Comment

                      Working...
                      X