Sneak Peek at WebAssembly Studio | Hacker News Reader

Sneak Peek at WebAssembly Studio

hacks.mozilla.org

192 pointsbigato40 comments

This looks awesome. Really useful tooling + gorgeous interface + clean fonts. Mozilla's looking pretty good lately.

The editor is from Microsoft https://github.com/Microsoft/monaco-editor

First the tor announcement and now this. That's a lot of exciting News from Mozilla recently.

Well, this was a couple months ago. See https://news.ycombinator.com/item?id=16811721 for the past discussion. Definitely still exciting though.

Fast forward few years later - web is made of executables and openness and user control is gone.

To be honest minimised JavaScript code isn’t much more readable.

This is why Stallman says we need to push to have a link out to the un-minified (or in this case source code that led to the WASM) so that people can go and read the source.

We already have this, there's nothing stopping you from using source maps in production. They're automatically loaded by browsers and can contain the full source code.

That’s a big downside and something I don’t like. But if we get actual universal bytecode out of it, it will have been worth it. Every platform out there, with the same bytecode and with a myriad of OSS languages out there able to target everything under the sun. It would take us out of the dark ages of having everything in C cause lowest common denominator...

1995 wants its Java 1.0 back ;)

Java doesn’t run on iOS.

Java doesn’t run on many, many other devices (which do support web techs).

Java is controlled by Oracle.

WASM is not the same thing.

Administrative and human factors matter.

You missed that I'm not comparing existing Java to WebAssembly?

Promises and hopes around Java 1.0 are very similar to WebAssembly.

Java was promoted mainly by 1 company. Web techs are made in agreement by a consortium of Mozilla, Apple, Google, Microsoft.

This has a way better long term shot at making it.

Maybe this one has better shot at making it.

I beg to differ about 1 company though (article from 1998): https://www.javaworld.com/article/2076748/java-web-developme...

Honestly the whole thing is like a bizarro Java world. Everything is significantly worse than the comparative technologies, some of which are 20+ years old. package.json is worse than pom.xml, build.ts is a bizarro build.gradle, and the vm itself doesn't even begin to compare. There's some kind of market failure going on when browser providers are devoting millions of dollars to poorly reinvent the wheel. On Android we'll be seeing 3 VMs running on top of each other. It's insanity. Frankly it does not inspire confidence in the future of browsers as a platform.

The VM may or may not compare, from some abstract standpoint, but:

1. the startup time and upfront memory usage of V8 and other JS VMs is incomparably lower (better) than that of HotSpot; and

2. Java in the browser was relegated to gray opaque squares, while JS / Wasm has control of the entire viewport and all browsers APIs.

Those two points alone make it a clear winner. Not to mention the horrible security track of Java, probably due to the complexity of its implementation.

#1 Java's poor startup time used to be a problem.

Even back in the day, the correct answer was Sun's HotJava, running applets within a native Java runtime, which was killed in the crib to appease NetScape.

You've heard the cliche about how sufficiently complicated 'C' programs recreate LISP, only poorly?

20 years recreating WORM applets. 20 years recreating grid computing and software agents (now pimped as 'serverless').

Imagine the world today if Sun had had some saavy championing their own tech.

#2 Canvas. The inevitable omega point will be tunneling X-Windows thru the HTTP, HTML, CSS, JavaScript (WASM) obfuscation & error injection gauntlet.

> On Android we'll be seeing 3 VMs running on top of each other. It's insanity.

Why does it follow that development of abstractions will lead to people stacking endless VMs on top of each other? Javascript runs on Android today in a native VM, not a java-based one. Why would that change in the future?

Java would have really rocked the client side web world if Sun had figured out the interface between DOM and Java applets instead of pushing their GUI implementation.

Just like when the entire web was Flash and Silverlight?

Choose HTML5. Choose Node. Choose your fucking social networking sites. Choose open APIs and a per-year subscription model. Choose microformats, micropayments, and SOA. Choose targeted mobile advertising. Choose user-contributed content. Choose your widgets in a range of matching Bootstrap colors. Choose a CDN, or two, or three. Choose the long-tail and wondering who the fuck you are on a Sunday morning. Choose surfing on that walled-garden device, a machine you paid for but do not own, watching mind-numbing, spirit-crushing videos, stuffing fucking junk food into your mouth. Choose rotting away at the end of it all, pissing your last in a miserable home, nothing more than an embarrassment to the selfish, fucked up followers you spawned to replace yourself. Choose your future. Choose WebAssembly.

(Not my idea, I just updated it for the times. Please don't mod down for foul language kthx, it's a homage to the original.)

What is this? Web Assembly (from the parent comment) will not be a choice for users right?

It's along the lines of the opening to "Trainspotting" a famous novel and film [1]

[1] https://en.m.wikipedia.org/wiki/Trainspotting_(film)

Ahhh thanks, I remember now!

Way to go Mozilla. This is awesome.

Binary explorer in cloud IDE is beautiful ;)

I felt like seeing a lisp machine

(what (really (and (only (matters (is (readability)))))))

    (-> readability is matters only and really what)

Like so?

https://clojure.org/guides/threading_macros

yeah, web assembly vm is a stack machine

not only that, the fact that you have programmatic access to low level semantics up to presentation into a "single" window and language. Very dynamic and multilayered

I've never seen a Lisp machine, but when I use Chrome's developer tools, which can display complex, interactive objects right in the text console, that's what I think about.

Devtools reminds me of the Smalltalk environment.

it's not far in spirit, but it's still quite far away

Nice video but (unrelated to the IDE) in order to use a printf from js it was necessary to: provide a js emulated syscall to c file that calls another c function that calls back to js again

Yeah, it’s super weird when most people are expecting this to compile into JavaScript running in a browser. Nevertheless, to be fair, this isn’t a C, C++, or Rust to JavaScript compiler-compiler it’s a C, C++, or Rust compiler that targets WebAssembly. I realize this might read as overly pedantic but it explains why it makes sense to the implement the C interface. It’s also a reasonable optimization pass for a mature compiler when it knows that you’re targeting WebAssembly running in a browser.

Out of curiosity, does anyone know what the sample_rate query parameter in this URL is for?

The WebAssembly Studio may track its usage through telemetry. We use this information to improve user experience. This information is not made publicly available.

This is very cool

     env: {
      __syscall0: function __syscall0(n) { return syscall(instance, n, []); },
      __syscall1: function __syscall1(n, a) { return syscall(instance, n, [a]); },
      __syscall2: function __syscall2(n, a, b) { return syscall(instance, n, [a, b]); },
      __syscall3: function __syscall3(n, a, b, c) { return syscall(instance, n, [a, b, c]); },
      __syscall4: function __syscall4(n, a, b, c, d) { return syscall(instance, n, [a, b, c, d]); },
      __syscall5: function __syscall5(n, a, b, c, d, e) { return syscall(instance, n, [a, b, c, d, e]); },
      __syscall6: function __syscall6(n, a, b, c, d, e, f) { return syscall(instance, n, [a, b, c, d, e, f]); },

Reply on news.ycombinator.com