This terrifies the linux shill

rust powered microkernels will destroy you
REDOX OS

Attached: 1920px-Redox_logo_2015.svg.png (1920x923, 55K)

Other urls found in this thread:

redox-os.org/
youtube.com/watch?v=HgtRAbE1nBM
twitter.com/SFWRedditGifs

Good luck, you will need it.

it has the drivers in the userspace

Oh no, competition in a collaborative environment, how horrifying, now people will compare and share ideas oh nooooooooo

what are you implying?

>lets shill a DOA OS so Microsoft cements its power in the Linux Foundation while developers are busy

Attached: 1254504784534.jpg (449x271, 28K)

> not using the one OS to rule them all

Attached: terminal-window.png (1884x1560, 129K)

> Written in a SJW language

This, Fuchsia, and Genode are things I hope keep developing further and become usable as a daily driver. Microkernels have progressed so much since Mach, Minix, Hurd, etc. They can compete in terms of performance now. All we need is for development and support to go into them, and with Redox having the novelty of being in a trendy lang, Fuchsia having Google dollars behind it, and Genode being built atop the highly-secure and military-supported seL4 kernel, I think it could happen.

no, it doesn't. it can, and perhaps someday will have drivers in userspace. right now you're lucky if you manage to use the limited VESA part of your GPU for display output.

I too have high hopes that one of the microkernel projects will amount to something usable. One "minor" problem with them in general is that it's not that hard to write a functioning microkernel compared to writing enough things around it to make it usable for.. well, anything.

You will find that once you have a fancy microkernel and you want to use it you'll have a desire to have things like a USB stack so you can use keyboards and mice. You'll probably want to use SSDs and HDDs and so on. And perhaps you'd like something to be output to a display? There is a slight difference between being able to have drivers running in user-space and actually have drivers.

im researching this and i went to the website

it says "Drivers run in Userspace"

redox-os.org/

The phone you're posting from runs on an SJW processor

This won't be desktop ready for another 10 years and by that time a desktop OS will just be a cloud service you'll access through ChromeOS.

rust is getting closer and closer to c performance, it already outperforms c++

im planning to help port kde, gnome, etc... to speed up things

Rust is only as fast as C or C++ when using unsafe code, right? C > unsafe Rust.
//Linked list in C
struct LinkedListNode {
struct LinkedListNode *next;
int value;
};

Show me a linked lists in Rust.

Tbh user, you need to use typedef to be able to make a variable of the type LinkedListNode inside itself.

>talks about performance
>brings up linked lists
is dis guy serious?

Nope, that works

It's a trade-off between safety and performance.
Nobody is saying that Rust is faster than C. And nobody is saying that C is safer than Rust.

Does it make sense to say C++ is slower than C when nearly all C code is valid C++? I don't see why the same code should be slower in a C++ compiler that uses the same optimiser anyway..

The recommended c++ idioms are supposed to be zero cost, so I'm curious as to why it should be slower.

The rate of development is pretty impressive given most is done by one BDFL.

hopefully. is it even remotely poosix though? If yes, Redox already failed.

>ambigiuous borrowing is not easily unraveled, therefor rust is unsafe!

or you could, just use nonshit data structures or do what any good C fag should do and prove out the correctness of an unsafe {} code block.

typedefing non-opaque structs is moronic. hell typedefs in general are cancer. They should only be used for abstract types of things you aren't supposed to consume directly.

as opposed to being not written at all?

i'd rather take some SJW, San Fran fart smelling faggot's shitcode than being a Jow Forums tard who doesn't do anything at all.

linux does not exist to make you feel superior to others, but to simply be good
if this redox thing turns out to be as great as you shill it to be then that's great news and there's no reason for people to be "terrified"

>this terrifies the linux shill
HAHAHAHOHOHO
no
even fucking templeos has more feature than this trash, it will never achieve feature parity with anything at their current dev pace.
maybe if you throw billions of dollar and millions of developers on it otherwise no, it will never replace anything.

>Rust is only as fast as C or C++ when using unsafe code, right?
yes and this is why asm/C will never die. There are things out there that can' be optimized or deduced by a compiler.
The current best way to do cutting-edge image filtering is asm with C/C++ bindings as compilers simply CAN'T optimize blocs that appears in morphology algorithms.
you should check x264 source code or the different av1 encoder/decoders or even try implement them in a fpga.
also when you develop a new ISA you have a new asm and the first pillar stone before adoption is having a C to asm compiler.
Once you have a C compiler you can run python, you can build a jvm and so on.
Then you have a C++ compiler and you can run everything that matter in this world.
You can always port erlang/otp if needed.
There is no asm without C for technical reasons and there is not high-level languages without C.

They can always try to delete C from existence but a similar language with the same features and risks will ultimately appears as you just CAN'T do without it.

youtube.com/watch?v=HgtRAbE1nBM
I know that this will activate Jow Forums's C autism but it's a fantastic talk and is worth watching for the history lesson alone

From what I know, it's partly posix. I think they're following posix for the things they agree with in it, but splitting away from it in areas they want to do differently.
Their ion shell language syntax for example is very different.

I guess that is what people said about Linux 25 years ago. People never learn.

Oh yes, I'm so scared of operating systems that need at least three context switches per syscall in an age where people are starting to run things in the kernel just to avoid context switching even once.
A world where NT is still faster than any *nix because it can do things in one highly configurable system call that *nixes need 5 for.

Microkernels are the kernel equivalent of something like JavaScript.
"Hey, computers have gotten faster, so let's pay a hefty performance tax for some convenience. Let's make our 16GB ghz machines run like 128MB 1.6ghz machines! Yahoo!"

>idioms are supposed to be zero cost
There is the hint.

Ever heard of Meltdown? Patching that made a good microkernel about as fast as a monolithic OS.

Have you ever heard about free lists? They are useful. I tried making an entity system in rust where a freelist would keep track of destroyed entities. However rust didn't allow me to cast a struct to int, and there were problems with serialization-traits that preventing me from casting struct to int. I don't remember the details, and I might just not have been experienced enough in rust.

I did use indices instead of pointers for the free list so borrow checking was not a problem, but my example code still demonstrates how limiting rust is, and I wanted to see a "good" linked list in rust out of curiosity. Also the borrow checker won't prevent bugs if you are using indices of of an array instead of pointers which makes "safe" code less safe than you think.
This guy knows his stuff.

Well there is nothing wrong using unsafe if you really need it. And no that doesn't defeat the purpose of Rust. If you have millions of lines of codes, yet only a few hundred of them unsafe that greatly reduces the probability of memory related bugs. Optional unsafety instead of the optional safety you have in C++.

The standard library is also still growing. There are new APIs to safely do things that were previously unsafe, or to do unsafe things more idiomatically.

>Hey, computers have gotten faster, so let's pay a hefty performance tax for some convenience. Let's make our 16GB ghz machines run like 128MB 1.6ghz machines! Yahoo!

But where is your proof moron?

what's wrong with posix? you fags shill plan9 like it's 2nd coming of christ.

posix-like. Perhaps it would be better to say inspired. It's not compatible but I doubt anyone using a *nix system would have a hard time adjusting.

>system language that doesn't even have any formal specification
why do rustfags expect anyone serious to actually use it? once it's standardized we can talk.

I do not shill plan9 as it's even worse than most poosix systems out there.

I think you're forgetting it took C over a decade to get a standard specification.
Rust 1.0 hasn't been out for 4 years.

>And nobody is saying that C is safer than Rust.
Lots of C programmers are saying exactly that.
I mean, they also say concurrency is an evil fad, so take that for what it's worth.

not an excuse. languages trying to replace C need to be at least as mature as C. mozilla should have kept quiet about rust until it was ready.

well, that's why people still use it. businesses won't even consider rust if there's no specification or standard and neither will i.

That's stupid.
You don't get the free labor, ideas, testers and people creating an ecosystem of libraries
Iterative design rather than an update every few years allows the language to evolve faster. This is especially necessary since Rust uses concepts not really used in other languages before that need fleshing out, that's why nightly exists.

Nothing is going to happen from it not being standardized yet.

>Rust 1.0 hasn't been out for 4 years
Standards and reference implementations are not the same thing. As much as I loathe sepples, at least there's more than one compiler for it.

Nothing bad will happen from letting Rust evolve naturally for a few more years before we need a standard. Like I said, not even C had a standard starting from its release.
It took 6 years from the birth of C until the release of 'K&R C', then another 11 until we had ANSI C.

you also don't get wide adoption until the language doesn't change significantly between small version updates. Probably part of the reaspn go is doing better in this respect. I like both languages though.

But Rust has a stable API. Any program written in 1.0 rust will compile with any 1.x compiler.

rust is not competing with early C. it's competing with modern C, with all the tools of modern C, and programmers who don't need training wheels for pointer arithmetic.

Imagine being such a terrible, terrible kernel architecture that you have to rely on a critical CPU bugs to affect your competitors to become relevant.

Almost. There are things you can do in 1.0 that weren't intentionally allowed, and have become errors in later versions.

>mit licensed
its bad because it denies your freedom but go off