Re:A first look at WebAssembly performance
by wyldfire on Thursday February 09, 2017 @10:14AM
(#13607450)
The promise that WASM holds IMO is for portability, not (merely) performance. How cool is it that I could write a game or application or library in C/C++/Rust/golang/swift/etc and expect it to be able to run on any modern CPU but also a browser?
[ Parent
| on HN ]
Re:A first look at WebAssembly performance
by zigzigzag on Friday February 10, 2017 @06:38AM
(#13614686)
I wouldn't say that's "cool", that's just what browsers could do 20 years ago with Java applets and ActiveX. Lots of languages target the JVM and you can compile C to it as well if you're willing to treat memory as a large byte array (there is a project that runs JIT-compiles LLVM bytecode on the JVM even).
The article shows that WebAssembly implementations often struggle to match Java. I see no reason to believe that browsers are actually easier to secure than a JVM, especially Firefox which doesn't even really do renderer sandboxing. They seem about the same level of difficulty. All the browser guys have done here is reinvent the same concepts of 20 years ago with a different instruction set that has "Web" in the name.