Rust 1.0 Is Coming Next Year

[phoronix]

Phoronix: Rust 1.0 Is Coming Next Year

The Rust Programming Language Blog has shared their plans for scheduling the Rust 1.0 release to happen next year...

http://www.phoronix.com/vr.php?view=MTg2MDk

[uid313]

Is anyone going to write a operating system kernel in Rust?

Would it be a sane idea and a technically sound decision?

Imagine if someone ported Linux to Rust...

[mark45]

Is anyone going to write a operating system kernel in Rust?

Would it be a sane idea and a technically sound decision?

Imagine if someone ported Linux to Rust...

Well these ideas are cool but doomed to fail like "let's write a killer app to kill Microsoft".
It's not practical to write a *new* kernel just because there's Rust, there are many languages like Rust including Vala and whatnot, should
we write a new kernel for each of them? Even if I could write a kernel, I'd have better things to do.

[ssokolow]

Well these ideas are cool but doomed to fail like "let's write a killer app to kill Microsoft".
It's not practical to write a *new* kernel just because there's Rust, there are many languages like Rust including Vala and whatnot, should
we write a new kernel for each of them? Even if I could write a kernel, I'd have better things to do.

Actually, Rust is pretty much unique in having a lot of similarities to C++11 but with compile-time enforcement of the C++ best practices for memory management and an ABI that allows foreign calls without needing the exact same compiler and headers to resolve the method name mangling.

Vala's syntax is horrendously crippled if you compile it in POSIX mode because the object model requires GObject as a backend (which makes it pretty much useless for systems programming).

Other languages with stronger memory protection than C or C++ like D, Go, and the various LISPs generally require some kind of runtime to manage memory which expects to run in user mode and sit on top of a kernel.

Ada could do it, but it's sort of the worst of both worlds in that it's more awkward than C++ and its protection model isn't as comprehensive as Rust's.

[uid313]

Well these ideas are cool but doomed to fail like "let's write a killer app to kill Microsoft".
It's not practical to write a *new* kernel just because there's Rust, there are many languages like Rust including Vala and whatnot, should
we write a new kernel for each of them? Even if I could write a kernel, I'd have better things to do.

Well you don't have to write a completely *new* kernel.
You could perhaps port the Linux kernel.
Perhaps it could be automated process.
Ximian automatically ported Android from Java to C# and .NET Framework.

Then you don't take over the world, you just switch the underlying kernel and keep on using the same userspace with the benefit of memory protection.

[xeekei]

Good luck getting Linus to take of his rose-tainted C glasses.

And good luck forcing every kernel contributor to learn Rust.

[uid313]

Good luck getting Linus to take of his rose-tainted C glasses.

And good luck forcing every kernel contributor to learn Rust.

Well, maybe a group ported it.
Then maybe everyone government and military started using the kernel based on Rust because it had less security vulnerabilities.

Then maybe distributions would start shipping the Rust-based kernel, and everyone pretty much abandoned the C-based kernel.

If the Rust-based kernel was superior and people adopted it, then maybe they would have to start coding against that.

[geearf]

Maybe instead of porting the whole kernel, you could make it accept modules in rust.

[JS987]

Then maybe everyone government and military started using the kernel based on Rust because it had less security vulnerabilities.

It's more like they will insist on using C language to keep it insecure.

[CrystalGamma]

Maybe instead of porting the whole kernel, you could make it accept modules in rust.

You can already export symbols for the C calling convention. One of the major roadblocks would be maintaining the Linux "headers" in Rust, as you cannot use the original headers.