LLVM's Clang Adds Initial Support For WebAssembly

[phoronix]

Phoronix: LLVM's Clang Adds Initial Support For WebAssembly

WebAssembly, the low-level programming language for in-browser, client-side scripting that's a joint effort by all leading web browser vendors, continues making progress...

http://www.phoronix.com/scan.php?page=news_item&px=WebAssembly-Initial-Clang

[Creak]

I'm really interested in this WebAssembly technology. I'm sure that being able to code a dynamic web site using C++ (my preferred language), or any other languages, is going to revolutionize web development. My hope is that, in the end, we're going to get rid of all this pile of web technologies (HTML, CSS, Javascript) that are so not practical that we have to create layers of abstraction above them.

My only concern is more about the fact that web sites will be less opened because WebAssembly will obfuscate the code and no one will be able to read the code anymore. But the reality is that some companies already obfuscate Javascript scripts (officially "because it's faster") and from what I know about free software, it is that it's useless to force a company to be opened, it has to understand the benefits by itself.

[Szzz]

My hope is that, in the end, we're going to get rid of all this pile of web technologies (HTML, CSS, Javascript) that are so not practical that we have to create layers of abstraction above them.

What do you propose instead of HTML? Directly render to canvas? How then search engines will index that?

[nils_]

I'd also be a lot more weary of malicious web sites breaking out of the browser sandbox. There is already too much code being run.

[Creak]

OK, just bear with me for a few minutes

My point is that, originally, the HTML language was willingly too simple so that anyone can create a page on the internet, even someone without any knowledge of computers. While the idea is commendable, the reality is that it didn't work. Building a web site still requires a lot of computer skills, to the point where people (even technical ones) prefers to use services like Facebook, Wordpress, Medium, etc. to create their pages.

So now that we know that, I think we should learn from that and move on. What we need is a stricter way to present data. More and more, I see websites using JSON (or XML) to get data from their database and present them using Javascript. This is not really search engines friendly too, but that's what it takes to get a dynamic and efficient website nowadays. Following this trend, I feel that the age of using a server side language (e.g. PHP) to fetch data and present them in simple HTML is about to end. Server side languages will still be useful, but to answer some web service requests.

I also think CSS is too complex for its purpose. It's even a joke among web developers that it is so difficult to center a text vertically. It's like the most basic thing you'd like to do and it's needs you to understand everything about the dataflow of your browser.

Based on that, I think we need to rationalize web development. HTML is *a way* to present data, but definitely not *the best way* of doing it. It's still possible to present data using a more structured language like JSON for instance. But there may be a unknown way of presenting it yet. Anyway, technically, it's not impossible nor that ludicrous to want to replace HTML.

[Creak]

I'd also be a lot more weary of malicious web sites breaking out of the browser sandbox. There is already too much code being run.

Since the code will be sandboxed into the browser, I don't think security is more a problem than it is today.

[Marc Driftmeyer]

HTML 5.x/6.x CSS 1/2/3/4... are here to stay. Deal with them or don't be a front end developer.

[Daktyl198]

If I hate to write a web page using JSON, I'd kill myself. So much unnecessary crap that's great for data structures, not so great for visual structures. HTML itself is okay, I think CSS is the real culprit here. CSS doesn't seem to be nearly as "standardized" as HTML does, with each browser having default margins, font sizes, etc set to their own preferences. Not to mention how fucked it is when you get a semi-complicated page going, and you change a value on one tag, and have a tag halfway across the page (unrelated to the first one at all) decide to suddenly break...

[Creak]

HTML 5.x/6.x CSS 1/2/3/4... are here to stay. Deal with them or don't be a front end developer.

I'm sorry, but that's not the point. I'm talking about possible evolutions of the web technologies and you tell me to deal with present ones... I agree that right now I have to deal with it, but it doesn't prevent me from having some opinions about it.

[kpedersen]

What do you propose instead of HTML? Directly render to canvas? How then search engines will index that?

I am pretty sure we *can* come up with a better solution than web crawlers scraping html documents.

It seems that HTML / CSS are adequate technologies for those who can stomach them (I personally can't) but the fact that pages from even the "big boys" like Google, Microsoft and Facebook occasionally appear broken on browsers (more often than not, mobile browsers) does suggest a certain amount of failure from these technologies. Sure, the developers are at immediate fault but I am pretty sure the current web standards aren't completely blameless for such breakages. Yes, backwards compatibility of legacy cruft is hard to solve, but drawing images to a canvas (i.e HTML5) is a much simpler operation and in my experience much less prone to breakage to any browser supporting html5.

So perhaps, HTML and CSS should be released as something like standalone libraries which render to the html5 canvas so developers can keep with a certain version and the browser can remain as dumb as possible to avoid legacy breakages.

EDIT:
My colleague next to me just suggested a global render farm with thousands of computers running IE6 and we just stream the pre-rendered pages from that... nice!