3 ms·

> To improve our critical infrastructure we must improve the developer experience (DX) of systems programming, but rewriting everything is not the only answer.

> To improve our critical infrastructure we must improve the developer experience (DX) of systems programming, but rewriting everything is not the only answer. I like Kristoff from what I've read of him. He seems like a leader who will take responsibility and not shy away from hard decisions. On that quality alone, my money is on Zig winning systems programming over Rust in 10 years.

xedrac 5y ago
When choosing between Zig and Rust for systems programming, I would reach for Rust most of the time. It's just a more expressive language with some seriously awesome compile-time guarantees. There are three things that bother me about Rust: bloated binaries, slow compile times, and dependency bloat (accidentally building 5 versions of the same crate). But seeing as it's being integrated into the Linux Kernel, I think Rust has a long and bright future in that space. Having said that, I think Zig would be a really great language for kernel programming.
0xCMP 5y ago
You can build C projects using Rust and Cargo as well[0], but it requires more work than Zig appears to need. Rust also has good C FFI story and honestly I think we will still see a lot of important code rewritten in Rust. That said, a more ergonomic C which seamlessly replaces the often bloated build configuration of C/C++ projects while allowing a far nicer language to slowly replace the existing code feels like TypeScript for JS (which I consider a good and smart thing). Definitely agree it's going to get far more use than I'd originally have guessed. [0]: https://adventures.michaelfbryan.com/posts/how-to-riir/ https://adventures.michaelfbryan.com/posts/how-to-riir/
- steveklabnik 5y ago
  This is qualitatively different, that requires a C compiler to be installed. rustc cannot compile C code itself. zig can.
roca 5y ago
In the 21st century, if you're building systems infrastructure and are able to choose a language that's not C or C++, it's a pretty big call to decline memory/type/datarace safety.

xedrac 5y ago

When choosing between Zig and Rust for systems programming, I would reach for Rust most of the time. It's just a more expressive language with some seriously awesome compile-time guarantees. There are three things that bother me about Rust: bloated binaries, slow compile times, and dependency bloat (accidentally building 5 versions of the same crate). But seeing as it's being integrated into the Linux Kernel, I think Rust has a long and bright future in that space. Having said that, I think Zig would be a really great language for kernel programming.

0xCMP 5y ago

You can build C projects using Rust and Cargo as well[0], but it requires more work than Zig appears to need. Rust also has good C FFI story and honestly I think we will still see a lot of important code rewritten in Rust. That said, a more ergonomic C which seamlessly replaces the often bloated build configuration of C/C++ projects while allowing a far nicer language to slowly replace the existing code feels like TypeScript for JS (which I consider a good and smart thing). Definitely agree it's going to get far more use than I'd originally have guessed. [0]: https://adventures.michaelfbryan.com/posts/how-to-riir/ https://adventures.michaelfbryan.com/posts/how-to-riir/

steveklabnik 5y ago
This is qualitatively different, that requires a C compiler to be installed. rustc cannot compile C code itself. zig can.

roca 5y ago

In the 21st century, if you're building systems infrastructure and are able to choose a language that's not C or C++, it's a pretty big call to decline memory/type/datarace safety.