pouët.net

Go to bottom

NaCl

category: code [glöplog]
http://code.google.com/chrome/nativeclient/
http://www.naclbox.com/gallery

Second Reality on the web anyone? Someone should do a website/service that lets you run old demos using DOSBox, WinUAE, etc...
added on the 2011-09-16 18:36:23 by mrdoob mrdoob
Chrome 14+ needed though.
added on the 2011-09-16 18:37:11 by mrdoob mrdoob
I really hope that PNaCl takes off and gets adopted by other browsers as well. Finally some sane browser coding environment !
added on the 2011-09-16 19:06:54 by xyz xyz
yes, let's all praise the evolution of activex!
xyz, in a certain way this sounds like Java all over again :) ... Then again, the LLVM bitcode isn't the worst thing to have as worldwide VM standard.
added on the 2011-09-16 19:35:52 by kb_ kb_
rasmus: Since I really don't know too much about ActiveX, could you be more precise? I hear this all the time, so I would like to know what is it that justifies that statement?
added on the 2011-09-16 21:33:44 by Hyde Hyde
Sodium under pacifiST under NaCLbox?
added on the 2011-09-16 22:01:21 by baah baah
yap sounds like java.
But is java that bad?
I don't think so...
Java is much, much faster then javascript in V8.
I think the main problems with java is loading time, using to much memory, and the garbage collector.
I hope NaCL would address in this issues...
Nacl is not activeX...

Nacl is much more like Java.

But Nacl suports C, and C++, and will have OpenGL ES support :)

I'm a Nacl fanboy :)

i'm curious about me minimum Nacl code....hum... is a 4k Nacl possible?!
Wouldn't it be great to have a central NaCl "platform" to see all those old intros without having to wire up yourself with dosbox and emulators? pouet staff? :)
added on the 2011-09-17 00:15:27 by xTr1m xTr1m
xTrqm, yap, i think that will happen.
Nacl is more like a in-browser VMWare actually.
Adlib tracker II can't run here. when I try to type imgmount d c:\at2.iso it just freezes :[ Any suggestions?
added on the 2011-09-17 00:24:33 by nula nula
how did I know that it would not take more than a couple of posts until someone mentioned Java :D

Well, as far as I have understood the goals of the PNaCl ("pinnacle") project, it is about providing an
open-source, secure VM / runtime environment that is not bound to any specific programming language,
does not enforce garbage collection, and provides basic interfaces for audio, opengl|es, input events,
(locked-down) local file access, and probably access to the browser DOM.

And yea, along with that you get platform agnostic (LLVM bytecode based) executables that are translated
to the actual platform ISA (x86 32bit/64bit, ARM, ..) "just in time".

I hope the google guys don't water this down by NOT keeping it simple (i.e. by complicating the HAL/default APIs)
or making it insecure (DOM/file/cookie/audio/GL/.. access for example should be user-controllable)

If they get this right, this could be the next big thing and everything I am hoping for in terms of
client-side web programming.

Flash/JavaScript alike middlewares can be built on top of it, and yea, you can even use a C/C++ compiler or assembly
if that's your thing.

The performance impact should be neglectable (for certain applications, hand-crafted X86 SIMD assembly will still
beat it but that it highly platform dependent anyway)

According to their roadmap, this thing should be ready in 2012.

@rasmus: This has nothing to do with ActiveX, which directly executes native code without any further checks.
added on the 2011-09-17 00:34:00 by xyz xyz
Works here.
i had to mount the iso manually (the iso filename in upper case).
xyz, yes i totally agree with you.

The OpenGl / graphic card problem will be hard to solve... :(
It's not only a Nacl problem, it's also a webGL problem, a silverlight problem, a java problem (with JOGL) and a flash 11 (Stage 3D, previously codenamed "Molehill") problem...
Graphics cards drivers where not build for secure computing... they were build with performance in mind...
The combination of possible hardware/drivers/OS it's huge!

Software rendering can not compete even with a low end graphic card... :(

I think intel larrabee was a nice answer to this problem, no more drivers.

But larrabee is dead :(
PauloFalcao doesnt work here, tho. latest chrome installed.

http://nooly.marshals.cz/temp/fail.jpg

Can you please tell me what am I doing wrong and how exactly should I mount that image?
added on the 2011-09-17 00:55:38 by nula nula
nula, try this:

imgmount d AT2.ISO -t iso
D:
adtrack2.exe
nobody said that it would be easy :) but if there is one company on this planet that has the resources and attitude to see this through properly, it's google, IMHO

According to this webgl wiki entry < webl security concerns > , DOS is their greatest concern.

that's one of the reasons why such features should be opt-in by the user and NEVER be enabled by default.
added on the 2011-09-17 01:08:12 by xyz xyz
"image file not found". now i'm lost :o. especially when "dir" command at C:\ shows "AT2.ISO" :[
added on the 2011-09-17 01:15:38 by nula nula
ok, tried it as well, got the same error.

"imgmount d AT2.ISO -t iso" does the trick

(all uppercase filename being the key difference - must be a bug)
added on the 2011-09-18 16:40:54 by xyz xyz
must be buggy, indeed :] yesterday it worked for me, today it doesn't work again (i'm sure i typed the same command : )

weird.
added on the 2011-09-18 20:05:56 by nula nula
Quote:
i'm curious about me minimum Nacl code....hum... is a 4k Nacl possible?!


I downloaded the SDK today and did some (small) tests. For the time being, it does not seem possible.
Infact, there's an open ticket about the (unnecessarily) large .nexe file sizes in the google bug tracker.

On the upside, I was pleasantly surprised to see that the performance of my test-app (no gfx output, just a simple prime number benchmark)
was really good (actually slightly faster than the native version but that's probably because the nacl gcc is based on gcc 4.4 and my
mingw installation only has 4.3)

I also spent a lot of time reading up on the security/VM subject.

Here's some links in case you are interested in the technical aspects:
PittSFIeld (software-based fault isolation) and
Overview of Native Client for ARM (low-level implementation details)

Well, personally I think that the web today is an absolute security and privacy disaster. Technology like this might fix the security aspect
but IMHO, whoever decided that client-side code execution (or scripting, for that matter) in a web-standard was a good idea was severely short-sighted.

The reason why I still think that NaCl is a good thing is that I believe that there should be a clear separation between the informational web
and the entertainment side of things (games, demos, etc..).

The amount of web-pages I encounter today that require scripting for what essentially is a static web page truly baffles me.

Hey, I know I know, this is pouet and this thread is even tagged off-topic, so I'm now gonna finish my wine and STFU :D
added on the 2011-09-19 00:17:18 by xyz xyz
so a thread that talks about a "new" demo platform is tagged off-topic? go pouet go!
added on the 2011-09-19 02:19:12 by mrdoob mrdoob
this thread makes me think of salt everytime.
added on the 2011-09-19 02:30:44 by wysiwtf wysiwtf

login

Go to top