4 ms·
The New JavaScript Engine in Internet Explorer 9
- elblanco 17y agoNow all that has to happen is to get this puppy on the majority of Windows boxes in the world.
- towndrunk 17y agoFirst, fix the CSS support so it's in line with all the other major browsers.
- daeken 17y agoI sincerely hope that they're using the JIT from .NET for this, although I've not seen .NET so much as mentioned anywhere near it. They already have a high-quality compiler that can emit insanely fast dynamic code, why not use it? Although honestly, I'd take it one step further: start rewriting everything possible in .NET, and improve the compiler in parallel (although they're obviously already doing this -- they're making good strides). If you limit the portions that are in unmanaged code to the absolute most performance critical components, you could honestly tout it as the most secure browser. Not to mention the benefits of their generational GC, or the development time improvements. If they could cut their development and debugging time in half, they could potentially build a browser that actually doesn't suck (which, sadly, I don't believe exists from anyone right now).
- ZeroGravitas 17y agoThe latest comment from 'ieblog' says that it's not the same code, though some of the same folk worked on it.
- thmz 17y agoI'm wondering: would it be possible for a compiler to check if parts of the javascript code could run on another core? For fun I created a 3D-render-engine in javascript + canvas (path tracing from lightsource). And I'm amazed how javascript speed differs in different browsers. Opera 10.50 and Chrome are like 3 times faster than others. But multi-threading would increase the speed even more (improving my program skills also...): www.thuijzer.nl/jsrender
- silvestrov 17y agoWeb Workers was born for this purpose. http://www.whatwg.org/specs/web-workers/current-work/ http://www.whatwg.org/specs/web-workers/current-work/ (and it would be non-trivial to detect this kind of code in Javascript).