6 ms·

He talks of a new memory safe kernel disrupting Linux the way Linux disrupted others. Well, maybe that's exactly what should happen. Maybe rust should have nev

He talks of a new memory safe kernel disrupting Linux the way Linux disrupted others. Well, maybe that's exactly what should happen. Maybe rust should have never been allowed in the kernel in the first place. It's caused a lot of trouble so far and for what? The dev community is fractured and infighting is rampant.

thinkharderdev 2y ago
> It's caused a lot of trouble so far and for what? The "for what" seems pretty obvious, no? It's the same reason you would use Rust over C in any other project. Memory safety and more modern PL features.
- superfist 2y ago
  In my opinion, in the case of the Linux kernel, Zig (once it reaches 1.0) would be a much better, more natural, and less painful choice than Rust.
  - capitol_ 2y ago
    Lets hope that zig becomes a memory-safe language before 1.0 then.
  - mrklol 2y ago
    Like stated, zig isn’t stable yet. And in general the kit isn’t on a good level yet to replace C in Linux. And, there’s still no good and fast allocator: https://github.com/ziglang/zig/issues/12484 https://github.com/ziglang/zig/issues/12484
    - flohofwoe 2y ago
      Kernel code shouldn't need a 'good and fast' allocator though, since preferably it shouldn't heap-allocate in the first place, or at least not frequent enough that allocator performance matters.
      - mrklol 2y ago
        You’re correct - was just something which is stopping people to try the language. And with that you implicate less users, less matured toolkit etc.
  - rowanG077 2y ago
    Is there even a plan to make zig memory safe?
  - bobajeff 2y ago
    As someone who's tried zig and who enjoys c. I highly doubt it would be any more acceptable for people dedicated to working on c than rust or c++ is.
stavros 2y ago
> Well, maybe that's exactly what should happen. Maybe rust should have never been allowed in the kernel in the first place. Can you clarify these two sentences? You say there should be a memory-safe kernel, and then you say Rust should never have been allowed, which is the opposite.
- PhilipRoman 2y ago
  That memory safe kernel doesn't necessarily have to be Linux itself.
  - simion314 2y ago
    Same opinion here, have all this giant proRust energy and put it behind a few Rust kernels, combining the Rust super powers with latest kernels architectures and have at least one Rust kernel that is not UNIX but something super modern.
    - lifty 2y ago
      There is Redox but adoption and traction is minimal. But these things take time so maybe it will become popular in the future.
      - simion314 2y ago
        I would have expected more then one Rust kernels. It kind of seems that Rust community is missing the ability or the desire to implement a kernel and would like the Linux kernel devs to do the work for them.
        steveklabnik 2y ago
        There are tons of embedded kernels, Android uses Rust in their Linux, Windows already ships Rust in their kernel. Just because someone mentions one project doesn’t mean it’s an exhaustive list.

thinkharderdev 2y ago

> It's caused a lot of trouble so far and for what? The "for what" seems pretty obvious, no? It's the same reason you would use Rust over C in any other project. Memory safety and more modern PL features.

superfist 2y ago
In my opinion, in the case of the Linux kernel, Zig (once it reaches 1.0) would be a much better, more natural, and less painful choice than Rust.
- capitol_ 2y ago
  Lets hope that zig becomes a memory-safe language before 1.0 then.
- mrklol 2y ago
  Like stated, zig isn’t stable yet. And in general the kit isn’t on a good level yet to replace C in Linux. And, there’s still no good and fast allocator: https://github.com/ziglang/zig/issues/12484 https://github.com/ziglang/zig/issues/12484
  - flohofwoe 2y ago
    Kernel code shouldn't need a 'good and fast' allocator though, since preferably it shouldn't heap-allocate in the first place, or at least not frequent enough that allocator performance matters.
    - mrklol 2y ago
      You’re correct - was just something which is stopping people to try the language. And with that you implicate less users, less matured toolkit etc.
- rowanG077 2y ago
  Is there even a plan to make zig memory safe?
- bobajeff 2y ago
  As someone who's tried zig and who enjoys c. I highly doubt it would be any more acceptable for people dedicated to working on c than rust or c++ is.

stavros 2y ago

> Well, maybe that's exactly what should happen. Maybe rust should have never been allowed in the kernel in the first place. Can you clarify these two sentences? You say there should be a memory-safe kernel, and then you say Rust should never have been allowed, which is the opposite.

PhilipRoman 2y ago
That memory safe kernel doesn't necessarily have to be Linux itself.
- simion314 2y ago
  Same opinion here, have all this giant proRust energy and put it behind a few Rust kernels, combining the Rust super powers with latest kernels architectures and have at least one Rust kernel that is not UNIX but something super modern.
  - lifty 2y ago
    There is Redox but adoption and traction is minimal. But these things take time so maybe it will become popular in the future.
    - simion314 2y ago
      I would have expected more then one Rust kernels. It kind of seems that Rust community is missing the ability or the desire to implement a kernel and would like the Linux kernel devs to do the work for them.
      - steveklabnik 2y ago
        There are tons of embedded kernels, Android uses Rust in their Linux, Windows already ships Rust in their kernel. Just because someone mentions one project doesn’t mean it’s an exhaustive list.