Poll: HTML5 or Native in 2 years?

[anigbrowl]

Which platform do you think will be more important in the future?

[melkisch]

The web is becoming more and more client side. Will native technology always be comparitively more powerful?

[mschuster91 → melkisch]

I think not, especially given the rise of JIT compiling. In 5 years we will be able to run JS at compiled-c-performance.

[kbenson → mschuster91]

I think we'll only ever have that for a specific subset of JS applications, where they are written in a way that lends themselves to very aggressive JIT optimizations. Other than those, my gut feeling is that there's too many high level abstractions that JS allows to expect it to get too close to C, who's abstractions map fairly closely to hardware.

[wslh → mschuster91]

I only expect that at least in 5 years we will choose the language and not be forced to develop in Javascript.

[millstone → mschuster91]

JIT compiling is hardly new, but JIT languages still underperform C. What makes you think that JavaScript will succeed where all others have failed?

[chc → millstone]

Lua says "Hi."

[millstone → melkisch]

Yes, because web browsers themselves are native apps, and so cannot do anything a native app cannot do.

[stray → melkisch]

of course

[chrisdevereux]

This isn't the right comparison. HTML5's competitor is native APIs, not native client code.

The question is really whether/when browser APIs, augmented by js libraries, will overtake native APIs.

My guess: 5 years, plus however long it takes for people to stop using whatever version IE is on by then.

[anigbrowl → chrisdevereux]

(I asked it on behalf of another HN user who wasn't able to access the poll interface)

[millstone → chrisdevereux]

Browser APIs must be built on native APIs. How can they ever overtake them?

[chc → millstone]

In the same way that C overtook assembly.

[suyash]

A mix of HTML5 and Native will emerge. Check out 'Native Client (NaCl)' and 'Portable Native Client (PNaCl)'

[d1plo1d → suyash]

WebGL strikes me as a much more likely and more open HTML5/Native hybridized future. I don't doubt there will be other hybrid technologies but I think WebGL is the first one that's here to stay.

To clarify: I mean open as in open web here; that is to say that the WebGL GLSL code that gets compiled to run on the GPU is human readable when embedded in a website. AFAIK this is not the the case with NaCL or PNaCL.

Moving back to adoption, NaCL has only currently been adoption by Chrome and Mozilla does not seem particularly interested in changing that.

Related:

* http://css.dzone.com/articles/mozilla-rejects-native-code

* http://news.ycombinator.com/item?id=4630057

[chc → suyash]

I'm not so sure about that. I thought these technologies sounded very intriguing early on, but I don't think they're going to have much impact on the future.

NaCl is weakly supported by Google, completely unsupported by other browsers and is totally architecture-dependent. It's quite possibly less portable than a normal native code app.

PNaCl is basically abandonware as far as I can tell.