Penguincrab: Unsafe Wrapper for Linux Kernel Library in Rust | Hacker News Reader

Penguincrab: Unsafe Wrapper for Linux Kernel Library in Rust

github.com

37 pointshomarp14 comments

Nice! System programming languages are having a revival.

I wonder if big companies will eventually rewrite their systems in those new languages. Will I live to see a Rust XNU/Linux/NT kernel?

What's certain is that the periphery is already being reworked - kernel modules in Rust, the shift to Swift, etc... Exciting times

Biggest mistake on Java and .NET languages was not having AOT compilation in the box like Delphi and VB 6 (to use common mainstream examples).

Had that been the case and probably we would never had the phase of JIT everywhere, followed by doing websites in scripting languages, to come full circle back to AOT compilation and first tier support for systems programing.

In any case, yes definitely exciting times.

We were all collectively drunk on Moore’s law.

Websites have head AOT-compiled components for ages, but ActiveX was an even bigger mistake than Javascript :)

That isn't what I was talking about.

Regarding ActiveX, ever heard of this thing called WebAssembly?

"USENIX Security '20 - Everything Old is New Again: Binary Security of WebAssembly"

https://www.youtube.com/watch?v=glL__xjviro

"USENIX Security '21 - Swivel: Hardening WebAssembly against Spectre"

https://www.youtube.com/watch?v=aLAHmnIkg24

WebAssembly is "AOT compiled" in the same sense that code for the JVM/CLR is "AOT compiled".

Read, my, comment, before, replying.

> That isn't what I was talking about.

A, few, more, comments, upwards.

> Biggest mistake on Java and .NET languages was not having AOT compilation in the box like Delphi and VB 6 (to use common mainstream examples).

I am thoroughly confused, because I did read your comments, apparently somehow differently than you read them.

You wrote:

> Had that been the case and probably we would never had the phase of JIT everywhere, followed by doing websites in scripting languages, to come full circle back to AOT compilation and first tier support for systems programing.

And then:

> Regarding ActiveX [the previous AOT compiled web thing], ever heard of this thing called WebAssembly?

You seem to be saying that WebAssembly is an AOT compilation thing that we're coming full circle to, and we're telling you it's not.

I was replying to the dude that mentioned ActiveX and Web development, even though my ORIGINAL comment made no reference to it, hence why I started my reply with "That isn't what I was talking about.", clear enough?

> I wonder if big companies will eventually rewrite their systems in those new languages. Will I live to see a Rust XNU/Linux/NT kernel?

I would be astonished to see any full rewrites; we're talking about things with millions of lines of code and decades of effort behind them. I could totally see incremental rewrites that start with a little Rust (or whatever) starting in one area and then eventually spreading one subsystem at a time (see: Linux offering the ability to write device drivers in Rust).

From what I understand of their product, Oxide computers (https://oxide.computer) is exactly doing this : building a server from the bottom up (OS, drivers and everything on top) in Rust. Anyone with more information can correct if I am wrong about their tech stack.

We are building those things, but to be clear about it, users workloads will run in VMs where users will have installed all the usual operating systems, we’re not like, reimplementing those or anything. We’re also using illumos. Servers have a lot going on!

Please excuse me for misrepresenting your work then Steve!

I take this opportunity to thank you for your involvement in the Rust project!

It’s not a misrepresentation, just not the full story. No worries. And you’re welcome!

Reply on news.ycombinator.com