MICROKERNELS DEFEAT MONOLITHIC KERNELS. END OF CONVERSATION

also,
>inb4 muh performance
Stop living in the 1980s. The world has moved on past mach. We live in a post-L4 world where microkernels can actually be competitive.
Also, if they were as unusably slow as FUD spreaders claim, Google wouldn't be making this.
en.wikipedia.org/wiki/Google_Fuchsia

Attached: Fuchsia.jpg (1268x664, 62K)

Attached: 4.png (313x403, 212K)

nobody seriously doubts microkernels, exokernels, or whatever-the-fuck-else-kernels are better. Are you going to build a kernel that uses that architecture? No. Is Linux going to be restructured to use that architecture? No. Why bother even discussing it?

you say that they defeat monolithic kernels however I can't help but notice that that overwhelmingly vast supermajority with only a statistical outlier saying otherwise uses monokernels

so no, no micro doesn't defeat jack shit

That's like saying everyone uses Facebook so that must mean it's good.

>he doesn't use facebook
why don't you want to interact with your fellow man
You're the weirdo for not using facebook

when the majority agree on something it's a pretty strong indicator that it's good, you're the weirdo for disagreeing

Attached: 1539792170555.jpg (845x637, 59K)

>Google wouldn't be making
And they didn't, that is a LittleKernel re-brand/fork.
github.com/littlekernel/lk

Microkernels would be great but giving end users more privileged access to the system is never a good thing. Layers of abstraction help protect users.

Yeah, it's a fork, and that's just the kernel. There's a userland and shit around that too.
>tfw an user falls this hard into the Linux kernel = OS maymay

Academic computer scientist here. Speak English, OP.

If that the case why not hybrid/modular kernel?

k

Do you need an ambulance for that stroke you're having?

Attached: 1536534936372.jpg (1280x720, 87K)

OP here I'm starting to get fucking mad lads, you will regret it.

Attached: 102155554898734.jpg (927x960, 80K)

nice image fgt

It's a technology board you fucking peasant.
Why not discuss kernel architecture? Or desktop thread is more up your alley?

I would rather prefer intel vs AMD thread over this fucking bullshit of yours

Micro images don't defeat macro images though.

What's the matter, the 50 existing consumer technology threads weren't enough for you? Fuck off brainlet. I'd rather see a thread discussing the theoretical possibilities of technology than your braindead consumerist bullshit.

I'm not OP.
And thanks for clarification. Instead discussion of pros and cons of technology, you would rather have endless gaymur manchild larping.

Muh product is better than yur product. Aaaahahaha. I gotz lik 15fps mor. I so superior.
Jesus fucking Christ

Attached: 1537650760911.jpg (465x465, 70K)

>being this brainlet
Intelfags need to consider suicide

Attached: 1497916694912.jpg (638x599, 123K)

CAN WE PLEASE, FOR ONCE, HAVE A LEGITIMATE DISCUSSION ON KERNEL TYPES AGAIN WITHOUT ALL THIS SHITPOSTING?

Seriously, why start out a thread like this and not want an actual chat about this topic like back in the day?

In most unix systems, there is one major component which one might classically expect in a kernel does in fact run in user-space, using the microkernel design. And that component is the X server.

When you are running a unix system as a server, for which a graphical environment is somewhere between an optional feature and completely unnecessary, this is absolutely an advantage. No discussion there.

But for pure desktop systems, is the user space X server an advantage? I'm not sure I would consider it one. It doesn't give me any real reliability advantages, because whenever the X server goes belly up, all my applications die with it. The distinction between an X server crash that needs a restart of X plus all user applications, and a kernel panic that needs a full reboot, is pretty academic to me -- which means the main theoretical advantage of this microkernel design aspect is mostly illusory in practice.

On the other hand, the user space aspect of the X server definitely costs me a lot in reliability. X has a bunch of fragile properties to its design that are a direct consequence of the fact that it lives as a separate program instead of a base operating system component; I am quite certain that a hypothetical X-like graphical component existing entirely as a linux kernel module would in practice be a bunch more reliable than actual X. No more userland binaries and libraries that need to line up exactly with the in-kernel drivers; no more hardware acceleration stuff that is configured correctly in the kernel module, and in a desktop app, but which is not supported by the X server because the userland part of the driver isn't configured properly against this particular version of the kernel driver.

All in all, in practice, I feel that the microkernel design aspect of X on unix is something that considerably reduces its reliability, rather than improving it. I am far from convinced that this would be any different in an overall microkernel-based operating system.

>literally using the abomination know as X.org to say that microkernels are awful
The current state of Jow Forums

harmful.cat-v.org/software/andy_tanenbaum
> - Microkernels are the way to go
> False unless your only goal is to get papers published.
> Plan 9's kernel is a fraction of the size of any microkernel
> we know and offers more functionality and comparable
> or often better performance.

>Plan 9
From Outer Space

I can understand why some would need to move device drivers and IPC to userspace, but why would anybody need to move filesystems, internet stack and some other very sensitive components to userspace? this is like getting liver or kidney out of the body and making it still work

>cat-v.org
The idiots who find their own existence bloated and harmful.

he is right, retard. If we ended up with all major components in userspace too like filesystems, networking, etc..., any failure in these userspace servers will likely cause a failure a system-wide failure since the clients will be affected to the extent that they won't even be able to communicate with these servers, and we will end up with the same fucking issue of the in-kernel components

It's fucking 26 years old mail.
>Name a product that succeeds by running UNIX as an application.
Cygwin.

check'd and kek'd

>Cygwin
kek

>harmful.cat-v.org
I think microkernels are a blast from the past too, especially considering how modern kernels are modular and hybrids, but you're literally posting a link to a satirical and deliberately inflammatory website.

Attached: uriel1.png (853x621, 87K)

buzzword bingo!

Haha, just imagine the time spent for syncing messages between modules, microkernels really suffer with deadlocks and starvations all the time.