/tech/ - Technology★

>>914711

so he will finally implement networking in templeos? i'll fucking switch to it if that's the case. it might be the most secure piece of software ever made. nobody would be autistic enough to write a virus in HolyC (except Terry, but he is a good person).

>web dev version of temple OS

>>914719

>t. brainlet larping as tech savvy

>>914708 (OP)

Interesting software....if it was of network.

You've got to be insane to shill web assembly as secure same thing for rust.

The only way to have something close to secure software is to have secure hardware which is something that does not exist (beside talos II).

>>914719

>obody would be autistic enough to write a virus in HolyC

Except other schizophrenic people who took control over temple os.

>>914723

Web assembly is shilled by google and MS their project is to migrate every function a OS as into the cloud.tm.

>>914728

B-but anon, why would you keep your files on your hard drive? They are protected from theft and corruption on the cloud(tm)! Also hard drives and sdds will be antics in the future! Why would you use such an old technology? Are you a pedo, anon? Or a terrorist? Why would you have something to hide?

>>914729

>I-Is Rust too hard for you, a-anon?

I do vhdl everyday, take your over abstracted shit safe space tier meme language somewhere else.

>>914719

He talks a little about it in that video. His concept is something completely separate from the Internet tnough, not something you'd use to browse websites on.

As far as virus go, those weren't such a big deal in the old days when you didn't have lots of devices vulnerable to firmware exploit. I once got the Stoned virus on my Amiga 500. Big deal, I simply shut off my computer before using any other floppy disks. It was pretty easy to keep nasty things contained that way. And if I had a virus scanner, it would have probably cleaned that floppy, but those days I didn't even have a modem. It was a game disk someone gave me, which was my usual method to get software besides buying it from the store. But the store didn't have everything, so...

How the fuck does webassembly even work?

Is it compatible across different architectures with different assembly?

>>914755

>How the fuck does webassembly even work?

Just like javascript.

>Is it compatible across different architectures with different assembly?

Yes, just like javascript.

Webassembly is just javascript Electric Boogaloo but with even more intention of removing the web and control over other peoples computer.

>>914767

If you want microkernels see:

http://www.draketo.de/english/free-software/howto-hurd-140-chars

>>914794

hurd is dead, anon..

>Mach is a so-called first generation microkernel. It is the microkernel currently used by the Hurd.

>only NFSv2 is supported so far.

>The Hurd presently has no sound support

>The DDE environment allows for reusing half of the Linux 2.6.32 network device drivers

>2.6.32

>support for character devices (like sound cards) and other hardware (USB, multicore) is mostly missing.

>Copyright © 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2013, 2015, 2016

>Last edited 2016

>>914767

furfags out

What is the point of this? If you want to be portable, just distribute the source code.

>>914719

There's fork of it with networking: https://www.youtube.com/watch?v=3oD8tj3eBls

>>914708 (OP)

>WebAssembly is designed to run safely on remote computers

Except it is still vulnerable to Spectre-class exploits for private data exfiltration, and now you've moved everything into the same address space so all of the data in working memory is available. Nice toy, but it's got 20+ years of processors that it shouldn't be run on.

>>914708 (OP)

If it weren't for side-channel attacks this would be a good idea.

WebAssembly is a synthesis of PNaCl (Portable Native Client), a Google project to run native code in Chrome, and asm.js, a subset of JS that can be compiled to native code. It's never been described this way officially, but that's how it evolved.

PNaCl was designed with security in mind. NaCl was a subset of x86 (if I recall correctly) that was sandboxable, and I think PNaCl was the same for LLVM IR. WebAssembly is similarly designed so that it can safely be compiled to native code. The safety comes from the limited and precisely defined semantics of WebAssembly.

Apart from correctness of the implementation of WebAssembly there are two other problems. The minor problem is that it is easier to target CPU bugs because WebAssembly will map closely to CPU instructions (it was designed to) so someone who knows a CPU bug could invoke it with WebAssembly. I thought I saw this done in practice but I couldn't find the link.

The major problem is side channel attacks. I don't see how you can avoid them without some kind of hardware support. Spectre/Meltdown expose weaknesses in the hardware isolation of processes, but nebulet won't use any hardware isolation, so it will have to do all the work of prevent side channel attacks in software. Since people have relied on (imperfect) hardware isolation for so long I doubt nebulet will be able to solve this problem on their own.

>>914733

>antics

You mean "antiques". Also you are officially too stupid to post on /tech/, please leave now

This actually sounds interesting, worth the developments. Is this probably how distributed OS may look like in the future? Since everyone and their dog is fatally attached to the concept of WEB APP? Maybe it will flop so hard people will actually want to try something simple, like Inferno.

but why

applications ... on THE WEB

>>914823

>>914719

Whoops meant to link https://github.com/minexew/Shrine

>it took websoylent to popularize such an obvious idea like Microkernels

>rust

>wasm usage in real world

>implicating safety and efficiency

what a stink

>>914758

Good summary.

Javascript was a mistake.

Webassembly is that mistake, fully weaponized.

>>914869

"safely compiled to native code"

Fucking hilarious.

Tohsaka Rin pleases young boys for free.

>>914719

You're advocating for security through obscurity. Stop that.

>>915111

More like security through simplicity and smallest possible codebase. Basically the opposite of "modern" design where they pile on more and more code to deal with inherent design issues, including ones in the C language itself.

But that said, Terry isn't going for TCP/IP. He wants something a lot simpler.

UTTERLY... FUCKING...

degenerate

nice, WASM in ring 0, just what we needed

>>mobile api in the kernel

DEGENERATE

I wondered when the singularity os concept would be revisited. Using web assembly allows them to reuse quite a bit of infrastructure, although it feels downright disgusting

>webassembly

>questionmark_nigger.jpg

what am i reading?