I know this will sound like I've been living in a server room for the past decade, but I'm going to go ahead and say it.
Does this feel backwards to anyone else?
We're building more and more on a layer that wasn't originally constructed to support more than text. I understand that the protocols are changing and that web browsers make a relatively great 'standardized' client platform. But in a world of portable, fast languages that can run anywhere, we still spin up our CPUs to turn HTML into buttons and animations with Javascript. There's something that seems hugely wasteful and contrived (hacky) about building more and more stuff on a mishmash of HTML + Javascript + CSS, when their origin (doing entirely different things) shines through so clearly sometiems.
I don't have a better answer, and I'm not even saying HTML5 is the wrong path, just pondering! Thick client, thin client, etc. mess, of course.
Edit: Upon further thought, it's the relatively high CPU use that bugs me primarily. Obviously, some JS engines are better than others, and I know theres a lot of work in that area. Plus, who's to say that JS CPU cycles aren't a legit performance cost for web apps?
It may be “backwards” when considered from a pure engineering perspective (in the same way as other large systems do – a government always seems “wastefully bureaucratic”, and any 500+ year-old city seems like a disorganized nightmare, for example), but I think that’s the wrong way to think about it.
The way the development of the web works is (1) the features that get used are the ones that are shipped (cf. http://diveintohtml5.org/past.html), and (2) the design and optimization decisions of browser vendors are shaped by the uses seen “in the wild”, and in general browser changes try to “pave the cowpaths”.
So the ultimate shape of the technical infrastructure is this mishmash of ad-hoc features that stuck around (and often were radically repurposed for uses well beyond original intentions), the accumulated experience of millions of websites and web authors, and the scars of occasional attempts to rationalize and coherently reorganize things. It doesn’t look pretty when you start peeking at the details, but it works because it slowly evolved from something else that also worked, all the way back to something simple.
Usually, attempts to supplant such evolved systems by “better designed” alternatives fail miserably, because even the most careful best-intentioned designers often miss some important arbitrary-seeming-but-actually-quite-important aspects of the original large messy systems. For instance, this is how I feel about most American city planning of the last 50–100 years, with its endless cookie-cutter suburbs and required automobiles. The goal was to simplify and rationalize and put everything on a coherent grid, and in some ways the new design is more “efficient” than the twisty streets and irregular hodgepodge of shops and apartments of some old-world metropolis. But quite a bit is also sacrificed with the new city plans, and it’s hard to quantify or even understand since many old-city advantages are based on idiosyncrasies and accidents.
Welcome to the word of organic evolution. Yes, it's messy and inefficient - because nobody knows beforehand how things would turn out. In hindsight, it's not rational to entrust junior programmers with single-handed development of initial versions of HTML and Javascript - but it was the only way these things could come out. Big, "rational", committee-driven projects (e.g. The Semantic Web) failed, and it's interesting to speculate why.
I wouldn't say that the Semantic Web has failed... it's moving slowly, but it's been adopted more and more. These sort of standards seem to have a critical mass kind of life... maybe compounded by the Internet hype machine.
Microframeworks for example have been very effective. I also hope we'll soon have something like FOAF spreading. Even Digg has made much of its context semantic now (check the source code)
I agree, but I'll go one step further. This whole time, I've pretty much been against the whole "web app" thing. I still prefer visiting (and definitely developing) websites that are basically just a bunch of hypertext documents—the kind of sites where URLs point to resources and don't change, HTTP methods make things happen, and (almost) every click of a link or button causes a synchronous request/response to/from the web server.
Now of course, there is at least a modicum of hypocrisy in my statement. A few "web apps" are nearly indispensable. Right now, Gmail is the only one I can think of, and the only "web app" out of the ~25 sites on my browser's bookmark toolbar. Everything indeed works perfectly, but I can't imagine how nasty the code behind it is (please, someone correct me if I'm wrong).
Lastly, I completely understand the main advantage of "web apps." It's just too darn convenient to deploy the things compared to any other platform.
Sure, I agree. I think the explanation is simply the usual "path of minimum resistance".
On the other hand I think HTML was meant for text and it still pretty much is. One thing I appreciate of the W3C is that rather than adding things on an existing standard they tend to create new standards. We have HTML, CSS, and SVG for example. CSS has also been modularised, and so was the plan for HTML. They've developed XPath, and MathML, RDF, and so on. For being a 20 years old format, HTML is surprisingly small.
The problem I think it's Javascript. But hopefully it shouldn't be too hard in the future to move to a language agnostic web.
My biggest complain at the moment would be one page apps, and some of ajax, that seem to break the old concept of the url. A small cost to pay though: I hope in the future they'll fix that.
Not really backwards but more like the wrong direction. For me HTML is all about displaying content. HTML then takes the form of an application when you can edit or interact with said content.
If we could start from scratch and build a set of tools for making web applications would they look like HTML and JavaScript? Probably not.
A lot of this looks great (new UI controls and css advancements in particular), but the multimedia elements seem very unpolished and as mentioned inefficient. The canvas demo even breaks the entire presentation if you scale the images larger than the canvas.
On simple internal projects for my company I often joke about 'post backs 4 life!' whenever someone mentions using Ajax or Flash for the UI. "Let's make it HTML5!" gives me that same sinking feeling.
Yeah, I particularly like the new range-slider input option, it seem much simpler to use than jQuery UI's range-slider. Which is where I think HTML5 really shines, making things simpler, by putting more functionality into HTML you previously needed JS for.
Does this feel backwards to anyone else?
We're building more and more on a layer that wasn't originally constructed to support more than text. I understand that the protocols are changing and that web browsers make a relatively great 'standardized' client platform. But in a world of portable, fast languages that can run anywhere, we still spin up our CPUs to turn HTML into buttons and animations with Javascript. There's something that seems hugely wasteful and contrived (hacky) about building more and more stuff on a mishmash of HTML + Javascript + CSS, when their origin (doing entirely different things) shines through so clearly sometiems.
I don't have a better answer, and I'm not even saying HTML5 is the wrong path, just pondering! Thick client, thin client, etc. mess, of course.
Edit: Upon further thought, it's the relatively high CPU use that bugs me primarily. Obviously, some JS engines are better than others, and I know theres a lot of work in that area. Plus, who's to say that JS CPU cycles aren't a legit performance cost for web apps?