The Web as a Platform? | Hacker News Reader

The Web as a Platform?

aalittle.com

38 pointsaalittle39 comments

The Web IS a platform. The things TFA is talking about, angular.js, Play, etc., are just tools / libraries. The platform is TCP, HTTP, HTML, CSS, Javascript, etc. And what makes it beautiful is that you can build anything you want on top of that platform, without worrying about what Google or Apple or Microsoft or IBM or $WHOEVER thinks.

Curation? Beautiful app-stores? Feh. I'll pass on both if it means giving up the freedom to build anything I can imagine, which is what you get with the general "web as a platform".

Never mind nobody is forcing you to "play keep up" and constantly use the very latest and trendiest tools all the time. I haven't even touched Angular.js, nor do I have any overriding desire to do so anytime soon. Same for Ember.js and any number of other libraries. I'm sure they're cool, but I'm too busy building stuff to spend all my time evaluating libraries. But it's neat to know that they are there, and one day down the road, if I start to feel the right kind of pain, maybe I look at pulling in one of those (or whatever the "library du jour" is at the time).

"I wish there was one straight-forward way to build awesome applications on the web like there is with mobile development. I want someone to hold my hand the whole time."

On the internet, you keep 100% of the revenue you bring in. On the internet, you do 100% of the work, too.

While you're right about 100% of the revenue (minus Paypal, Visa/Mastercard, etc.) I disagree you do 100% of the work when compared to iOS/Android development.

I actually love the tools we have available to us in webdev, and think it's actually quite a lot easier than native development for mobile platforms. It's so much easier to debug and work on a web app in real time in Chrome dev tools than it is to fire up the iPhone/Android simulator apps.

  > "quite a lot easier than native development for mobile
  > platforms"

I am going to bet that anyone claiming things like this did not ever half-heartedly attempted to look into native development and has no idea how much native frameworks have to offer.

You'd be wrong. I have done quite a bit. They have a lot to offer, but I think web development is easier and faster.

Still, you need to be way more qualified to build a successful SaaS or anything interesting on the internet. To build an iPhone app all you need is an Objective-C book, a mac, and $100.

All you need to build a web app is Chrome, a text editor and a $5/mo hosting account. :D

Maybe for you. Many highly used web services require non trivial backends. If you have an html5 app that uses local storage, you could host JavaScript for $5 a month, sure.

I agree, the plethora of platforms is a feature not a problem. The problem is when there is only one there is no innovation. The web is very healthy on innovation and iteration on how to do things better. In the end you build systems and products that work, using the best tools at the time, which isn't always the latest but sometimes is.

It's striking how often new web libraries come along. Are they genuinely better?

Very often yes, when you consider that the "better" question is not one-dimensional. Rails and Django are good at building traditional web apps where the logic and data live on the server. They have less to offer on the client side, where JS frameworks like Backbone, Angular, and Ember crop up. They also don't do async well out of the box, which you want if you're going to do streaming updates in real time. So node.js, EventMachine, and Gevent come along to solve that problem.

There are some very young frameworks (Meteor, Derby) trying to solve all the above problems at once. I haven't used them, but my impression is that they're not really there yet.

I think the underlying pain point from the article is that iOS and Android have the luxury of being able to iterate on the whole stack at the same time, which means they can move way more quickly than web standards, which can take years of argument and refinement, with disparate implementations in different browsers, which then get papered over in various ways by different client side libraries.

Yes!

Let's be honest here. If you had to chose between the web being gone forever and iOS+Android+whatever being gone forever, which would you chose?

If you want to be honest you should rephrase your question and use "all native apps on all platforms" instead of "iOS+Android+whatever".

And then you would understand, that question makes no sense.

Completely agree. The beauty of the web is it's plethora of choice (i.e., freedom). You decide what you want to make and how you want to make it. And nobody can stop you from doing it, or lock you down, or deny your request. You don't have to know every single tool to do good work either. But I feel a lot more comfortable on the web platform, knowing there is likely a solution to any problem I might encounter, than struggling to shoe-horn a single language/framework because that's simply all the platform-du-jour can support.

  > "than struggling to shoe-horn a single language/framework
  > because that's simply all the platform-du-jour
  > can support."

And in reality you just trying to poorly reimplement the stuff that platform-du-jour with decades of refinement has to offer. I don't want even to imagine what amount of effort is needed to replicate everyday stuff like Core Data + UITableViewController + UIFetchedResultsController and to make it even remotely performant.

> In some respects, the plethora of choice this problem represents is the power of the web, but overall I think it does the web a disservice because there isn't enough weight put behind a single platform.

Fair argument (I guess?) but the opinion comes down to, do you prefer an open and ever-changing ecosystem that has no limits on what you can build and how you can build it; or do you prefer a contained ecosystem dictated by a few powerful stakeholders who subtly apply their own methodologies and beliefs, to shape the capabilities and limits of the platform?

The latter seems like a slippery slope and would eventually serve the the powerful stakeholders rather than the people building the platform. The former, while it may require a higher learning curve, puts more competition between technologies that enhance the platform, and doesn't dictate what or how things can be built, serving the developers and users above all else.

Personally I would prefer the former.

Ahh... no, I'd rather that didn't happen. Not yet, anyway.

We are jumping around between client-side frameworks, but I think that's because a clear winner hasn't emerged yet. Each new one is reacting to the last, and the decisions are still being tested.

I think the situation will solve itself when the winner shows up. It's not just an adoption problem-- it's a utility problem-- so having a big company put their weight behind one choice would lock us in prematurely.

Lock-in ease. Most of the mobile frameworks attempt to lock you in to their platform. So they are specifically meant to be easy. This really is the Microsoft playbook of old. You want resistance and choice to be minimized in this case, and lock into tools and platforms due to their ease. On the web all the libs/frameworks are really just faster ways to use web technology (HTTP/TCP/etc), where native apps have to tie into hardware and systems so you must use them so there can be only one. There are actually lots of ways to develop for mobile beyond these frameworks though.

But even within these frameworks (Android, iOS) things are changed and swapped out at their will i.e. changing UIKit, Apple switching from Carbon to Cocoa, In-App Billing changing etc. So lock-in sometimes creates massive ripples of forced change.

If you are making cross platform apps, like the web is cross platform, that is where things get more difficult. Complexity goes up, more solutions appear to solve and iterate on the problem. The web and web tech is the best solution we have for cross platform currently.

Uhm no, thanks.

I have developed for Android and while it's XML-based layout system has nice enough foundations, it is really showing it's age. The interaction between different layouts, gravity hints, weights and screen sizes has grown nightmarish - and in contrast to the open web, there are not many good third-party additions to sanitize the morass.

You want your app to run on 2.2 - 4.1? You will have to re-implement a lot of the newer features yourself or stick with the old ways - something like jQuery that papers over the differences is rare on the single-provider platforms.

IMO the web as a platform had a lot of catching up to do but it's pace has outgrown both Android and iOS which are not exactly stale but do no longer innovate with the speed they did 4 years ago.

On the web, you can pick your tech stack and stick to it for some years, there's no need to switch everything over to the latest tech each year.

If you let Google choose, the platform will be in Java.

God, please no.

If you let Twitter choose, the platform will be in Scala.

God, please no.

That's why we are where we are. I don't want to develop in that gunk, but others do.

What would be the platform if we let _you_ choose?

I imagine Google would choose Dart for the web. They seem to have lost interest in GWT.

well, what's great about the web as a platform is that it is only one framework. the web. It renders HTML, styled with CSS, and is programmed using JavaScript. unless you count server side languages. you can use JavaScript libraries if you choose but that's your choice. If you decide to use Backbone.js and then you find out Angular is more popular, that's okay because in the end, you're still writing code for the same platform. that is why the web is so great. just write a good HTML 5 application and it will work on most devices--especially native html5 phones.

I don't think we'll see big companies throwing their weight behind iOS-quality web frameworks in the near term. It would be misaligned with their incentives. When Apple develops UIKit and the associated development tools, they convince developers to move to iOS. When Google develops Android, they convince developers to move to their platform. When someone makes a web framework, they convince people to move to a platform where they have no skin in the game. You don't sell more devices through convincing people to move to an open platform.

Further, I think the author hugely underestimates the effort that goes into designing an iOS/Android quality framework. Those frameworks go way beyond the MVC microframeworks that are the rage these days. These frameworks consist of many detailed high-quality images, gradients, sheens, and finely tuned animations. You can't just start an open source project and expect engineers to magically produce these assets in a coherent manner. It takes a large team of designers (and a dictator) many man-years to produce these assets.

This is the script.google.com and webshell.io project, making a cloud framework for APIs, so for the web-as-a-platform enabling developers to use webservices with a unique Javascript interface.

It follows the defacto standards of the web and not trying to invent a new standard.

- HTTP

- REST/JSON

- Node.js ( Google prefers using Rhino because Appengine does't support Node.js but Java)

- Github (Google prefers using Google Drive)

- Javascript as the glue of APIs by transforming API endpints into Javascript Objects, client-side or server-side.

- HTML views to generate with API calls

Webshell.io has an open API that Google Apps Script doesn't yet.

The webshell's blog is "building an internet operating system" for the web-as-aplatform, where APIs are the "drivers" of the web.

So definitively the web as a platform is coming.

The true question behind this post is : the web is a decentralized network, so will developers will use a centralized platform for making applications on the web? Is open source platforms the solution?

I don't see why that is a complaint or an issue? Its the good part about the web. No matter what you choose you can build your app using it. The standards around the web are open and with IE's popularity down you no longer have to learn the quirks of any specific implementation. So its the way things should be.

It's worth pointing out that there are a ton of frameworks and libs for propriety development environments too. Even where there is a dominant framework you then get micro architectures on top of that.

With the web as an application development platform it's still very young and there is yet to be a defacto standard for a few years. Until then frameworks will come and go. jQuery for example more or less won out against dojo or prototype (and I expect 100's of others that were less popular).

The democracy that drives the web forward is a double edged sword. While the Android, iOS SDKs are great now, they may well be dead in a few years, whilst the web will still be here. Just ask Swing, JavaFX, J2ME, Enyo, Symbian, Flash or Silverlight developers ;)

I have some sympathy for the author's complaint about a lot of web "platforms" compared to iOS or Android development. I have been going a little crazy for the last few weeks (http://blog.markwatson.com/2013/03/i-am-still-looking-for-a-...) trying different stacks for rich web app development for a new project. BTW, I settled on: HTML5 with Bootstrap with separate ClojureScript to manage all events and interaction with the server. After some of the more complicated things I have tried, this simple stack is like a breath of fresh air.

It's unfortunate that the typical web development experience forces application developers to shoulder most of the burden of assembling and release engineering their stack. Quite different from building against iOS, or Win32, or Android. That's too bad: it's a big time investment, and it limits the set of people who can build and maintain a real application in production to those who've learned a surprising amount of releng.

Imagine how different the Linux ecosystem would be without distributions. The individual modules don't all have to come from the same place. But I think the OP's point about curation is spot on.

That could only work if one company can produce something so comprehensively better for every task that it would be madness to choose anything else.

The only other way to achieve this would be for one cloud provider to gain critical mass and then drop support for anything other than their preferred framework.

For most people a safe and easy job is more important than freedom.

Can't blame anyone for this :)

The web is an open, diverse and ever-evolving ecosystem, and I hope that it remains this way for years to come. OK, there are no silver bullets nor panaceas, but that's more of a quality than a shortcoming.

Those companies are investing in that framework. It's currently in the form of a bunch of drafts on W3C. Web components and the shadow DOM are a few that embody what you're talking about.

As others have said, the Web is a platform and it's one that not just one company but all the big companies (Google, Apple, Mozilla, Microsoft, Yahoo!, Facebook, etc.) are all behind.

TCP, HTTP, HTML5, ECMAScript 6, CSS3. These technologies make up the Web Platform and it is consistent.

The problem is that rather than learning the standards, people like the author focus on learning libraries like Backbone.js or whatever and then when it's no longer the trend of the week they feel lost.

Learn the platform and you will be better off. You will find that picking up libraries is a lot easier when you understand what it is the libraries are built on.

Not a very good blog post, I must say.

ECMAScript 6? Now that's some serious optimism.

Ok maybe I should have said 5 but, yes I'm being uncharacteristically optimistic in hoping that learning it will be useful in the near future. :)

This reads like the classic frustrations of an inexperienced and overwhelmed web developer. The web is a platform. There are standard practices and toolchains. Many frameworks are backed by companies.

I enjoy web development much more than iOS or Android development. The freedom, the speed, the easy debugging, the numerous choices for the tech stack and development environment... The list goes on.

web is a "constantly patched" platform.. it was not created to do the things it does today, in mind.. and any good engineer knows that if you dont plan well ahead of time.. things will get scary over time..

the web is showing its age.. it time to move on.. is the application platform the answer? i dont think so.. but at least it show us some hidden truths about what a developer really wants and need..

the web platform was never a dream platform to us to embrace.. it was the only safe boat at a time, when platform wars were making us (technologists) more bald, poor and nuts.. you know.. because they always want the control.. markets, profits, whatever..

the thing to worry about.. its the: "ops, we did it again" factor .. now we got a dejavu feeling back to the nineties as a new platform wars arrives.. :(

I don't really understand what op means by 'platform' The example he gave are things of different category. Angular and ember are client side web frame work. Play is not. Dart is a programming language and WebKit is a browser engine.

Reply on news.ycombinator.com