Hacker News Reader

Comment by TuringTest | Hacker News Reader

Wait until mobile OSes start building a native presentation layer built on WebAssembly.

Apple already tried that approach with the iPhone one. It failed because the technology wasn't mature, but there have been nothing but advances since then.

The web is a more flexible and portable platform, so the economic incentives will always be there. And while native toolkits are still based on 1970's lab research, the web as a platform benefits from mainstream usage by the whole world. At some point, the new tech will overcome and replace the old approach. The gap is getting smaller every day even if the old approach has decades of advantage, and all major tech providers are pushing for it.

View on HN

With the security sandbox of the browser functionality is inherently limited (by design) compared to native apps which have less-limiting restrictions.

Security and usability are inversely proportional and the browser needs that security more than anything else.

That being said, it's great to see the innovation in the web space. But it will always be limited to specific application domains and the standards move slowly

Even by 2009 it worked fine -- I had a Palm Pre, and the experience of using webOS on that hardware was always responsive and pleasant. (Palm Inc. and the Pre both had lots of problems, but webOS wasn't one of them.)

In fact if I had the opportunity to buy a webOS device today I would ditch Android in a hot minute...

You can buy some webOS devices today: LG's "smart" televisions. It's actually quite slick.

XUL again?

If it's more convenient to Google and Apple, they'll push for it, and we can only adapt.

But don't expect developers to ask for it. As it happens with those things, each platform will have their own, incompatible toolkits, and programming for them will not be much different than native programming.

Why haven't they built a native UI toolkit for JS, though? What will WebAssembly give them that makes it worthwhile?

The ability to leverage more native capabilities than JS can. A more generalized, less platform specific format for executable code already seems to be a goal for iOS and Android. Android of course has always used Java (well, Dalvik technically) bytecode, and Apple recently has started pushing their LLVM-based bitcode to reduce dependence on specific architectures.

If WebAssembly does a good job of exposing features needed for high-performance code, it could become feasible to move everything to WebAssembly and ditch support for native code on both platforms.

WebAssembly doesn't make native integration any easier than JS. Neither defines the exact way so called non-web embeddings actually are defined and both can support it. I honestly have no idea the relevance of WebAssembly in this discussion. Could you provide an illustrative example?

Edit: I should clarify that for certain types of apps WebAssembly might be relevant but the discussion seemed to be out UI layer and I don't understand how WebAssembly fills a gap that a native API in JS could also provide along with the same challenges.

I'm sorry, I'm using "native" in an unclear way (that should have been obvious to me). I don't mean it will help with native API integration (JavaScript can do that just fine), I mean as a replacement for machine code on both platforms. iOS of course uses machine code for everything, but on Android it's mostly whittled down to the minimum: things that Java can't do efficiently, or code that isn't Java. That's the role I hope WebAssembly can fill, but it still remains to be seen how usable their solutions for threads, SIMD, etc. are.

A good example is games. Some do manage to use Java, but it's mostly a hindrance; being able to compile languages with a C style ontology to a platform-independent and efficient VM would be very helpful here.

> Wait until mobile OSes start building a native presentation layer built on WebAssembly.

Is there any active project you can link to regarding this?

WebAssembly is too new for this to be a thing yet.

>native presentation layer built on WebAssembly

Is your prediction that companies will start compiling their UI toolkits for browsers?

Not op but it seems like a logical step to at least attempt to do this, or something similar.

For the operating system makers?

Reply on news.ycombinator.com