I’m done with the web | Hacker News Reader

I’m done with the web

randyluecke.tumblr.com

376 pointspaulirish275 comments

> The community has tossed out everything we’ve learned from the last 30 years of building applications in favor of being able to put something together quickly with complete disregard for maintainability or extendability.

That's spot on. You can add in to that that all the lessons of the process of developing software have also been discarded and are slowly being re-learned.

It is as if with the birth of the web there was a complete reboot of the IT industry which accidentally also re-formatted the drive that held our knowledge about how you go about software development (which is different from knowledge about what it is that you are building, and different still from the technical details of your implementation).

On top of all that the web is incredibly messy.

> On top of all that the web is incredibly messy.

That is arguably both a bug and a feature. The web is an untended jungle, possibly the most diverse of any technical ecosystem.

Democracy and markets are messy too, but I think most of us would rather not give those up.

When it comes to technology messy is not a feature but a drawback. It precludes a very important thing from manifesting itself, reliability. The difference between the backbone of the internet (TCP/IP) compared to the maze that is web development is simply astounding. The amount of hoops you have to jump through to do stuff that was child's play on any computer 20+ years ago just because it is on the web completely baffles me.

People are actually proud of being able to make simple games or chat services using a browser. That is because it is so terribly difficult to do such simple things using the web as a delivery device.

- HTML - JavaScript - CSS - Server side code (pick your flavor) - HTTP - websockets - some real time extension (node or whatever) - a persistence layer

And all that for the basic functionality of tying a few people on graphics terminals to each other. Oh, and to make it more interesting we're shipping actual code from one location to another so you have a few security nightmares extra.

I think you're underestimating the complexity of native apps. To duplicate functionality of a web app, you're still going to need to know your networking layer, whatever it is, and you still need to write a server and persistence layer; the only difference is on the client, where instead of HTML/CSS/JS you're building in one language... plus probably a GUI framework (which are not exactly known for their simplicity or ease of use). And there are security issues related to distributing native binaries too. When you say there are tons of extra 'hoops' to jump through, what do you mean?

I was on the Internet regularly 25 years ago. "Simple network games" did not exist, they had incredibly complex protocols and failure conditions (remember setting up Netrek?).

Chat services also incredibly difficult to run correctly over the Internet that there was no proliferation, there was... IRC. Which had its own major messes (recalling EFnet netsplits almost hourly in the early 90s).

Or let's talk about BBS'. So, you could do ANSI art fairly easily, but let's recall the hassle of handling multiple users on the software of the day. You usually didn't. BBS games (Tradewars, Legend of the Red Dragon, etc.) were turn based - someone dialed in, played their turn for the day, and waited for everyone else to dial in and play their turn. Email and Fidonet EchoGroups came in daily, or maybe twice a day, when you dialed into your upstream provider. The only way you could run a multitasking BBS with multiple phone lines was mostly DESQview, OS/2, or the BBS software itself had its own multi-threading.

The Internet didn't replace IRC and BBS'... the Web did. Not that they're gone completely, just no where near as important as they once were.

Network games are OTOH are still as complex as Netrek because they don't fit a Hypermedia architecture.

But ultimately I think you have nostalgia for a better world that never existed. Which is fine if you want inspiration for new ideas but... it is important to understand why the Web won: for many cases, it was better than what we had.

While you've undoubtedly worked at some great shops with other great engineers...can you say that the IT and development industry isn't as rickety as web development seems today? I mean, how can we really know? Today, we have github and stackoverflow and a myriad of other coding boards in which we can see the most horrible of code snippets -- and the way the mind works, we usually remember the horrid examples more than we do the average or decent ones. But we lack an ability to perform similar introspection on older code bases. But I think we've seen enough shoddy legacy code to assume that development has always had a variance in quality

I've worked in banking, telco, cad/cam and embedded systems.

Nowhere, and I really mean that, was the tech as botched as it has been on the web past the early days.

To be fair, those are all systems with a drastically higher cost of failure than general web development.

Not to mention being developed in totally different environments than your average website, application, or framework.

I worked in a consultancy that primarily deals in websites and small apps for clients. Projects were generally underbid and handed to designers and engineers with the impetus to hurry your ass up so we can make money on this. As such, we were constantly throwing stuff together, duplicating code used elsewhere, no DRY, mounting technical debt. Don't get me wrong, this worked for us, so I couldn't complain much...

However, what irked me is that this mindset flowed into the SaaS and native mobile app-development side of the house as well. Fully or partly-spec'd native development is expected to result in something pretty and take no time at all, just like the web-design side. Very little interest in engineering things properly.

Fear of over-engineering absolutely should be ever-present as part of your efforts. But, I think we're taking this to extremes sometimes.

I avoided web for 15 years. Now I think it has matured and I am loving it! Cloud computing and decent frameworks with decent browsers.

It's more like the ease of Web programming opened up programming to an entire new slice of the population that would never have had the opportunity to become programmers before.

These people don't have computer science degrees, so they spend a lot of time reinventing wheels. But lots of it works well enough despite all that, and as time passes, they learn and get better.

The messiness is a good sign, not a bad one. Thirty years ago these people would have had no options other than to be passive consumers of software other people made. Now they're making things for themselves! And some of what they make is good enough to help other people with their problems, too. That's a positive development, even if the things they make aren't always pretty.

> Thirty years ago these people would have had no options other than to be passive consumers of software other people made.

Not exactly: https://groups.google.com/group/comp.sys.mac.hypercard/brows...

(That was 26 years ago, and HyperCard was far more accessible to non-computer scientists than the web is or has been).

It's a mistake to think that because the web is popular, that one can draw directly from it axioms regarding how technology should be implemented.

I'm old enough to have used HyperCard, and while it was an excellent environment, it had some serious barriers to entry. For instance, using it required an expensive development kit, namely a Mac. Back in 1987 even a lowly Mac Plus would cost you $2,500, which in today's dollars is more than $5,000. Contrast that to the Web, which you can author for on the oldest, crustiest commodity hardware you please and where even back in the mid-'90s you could get started with just a text editor and a few bucks a month for hosting.

"HyperCard was far more accessible to non-computer scientists than the web is or has been."

A simple comparison of the total number of HyperCard stacks built versus the total number of Web sites built would argue otherwise. You can argue about whether HyperCard authoring is more intuitive than HTML authoring (I think it was), but as soon as the Web came along people decided for themselves which of the two they preferred to hack on.

> I'm old enough to have used HyperCard, and while it was an excellent environment, it had some serious barriers to entry. For instance, using it required an expensive development kit, namely a Mac. Back in 1987 even a lowly Mac Plus would cost you $2,500, which in today's dollars is more than $5,000. Contrast that to the Web, which you can author for on the oldest, crustiest commodity hardware you please and where even back in the mid-'90s you could get started with just a text editor and a few bucks a month for hosting.

Any computer was not cheap in 1987, much less one that sported a usable GUI environment that could run something on par with HyperCard. That had nothing to do with HyperCard.

> A simple comparison of the total number of HyperCard stacks built versus the total number of Web sites built would argue otherwise.

There were far, far fewer people actually running computers, much less collaborating on the internet, 26 years ago.

> You can argue about whether HyperCard authoring is more intuitive than HTML authoring (I think it was), but as soon as the Web came along people decided for themselves which of the two they preferred to hack on.

When the web came along, Apple had already made HyperCard inaccessible by charging a significant amount of money for the authoring tools, rather than shipping it with each Mac.

Even still, people were producing "My Favorite Links" and home pages, not apps.

"Any computer was not cheap in 1987, much less one that sported a usable GUI environment that could run something on par with HyperCard. That had nothing to do with HyperCard."

True, but (1) PCs were significantly cheaper than Macs, and (2) many, many more people already had PCs on their desks than had Macs. So unless you were one of the few who already had a Mac, getting started with HyperCard meant plunking down money for new hardware.

"There were far, far fewer people actually running computers, much less collaborating on the internet, 26 years ago."

Again true, but irrelevant. We're talking about contemporary mindshare, and starting in the early '90s HyperCard and the Web were competing against each other in real time. HyperCard had a long head start and yet even by 1995 (the year Netscape 1.0 broke out) it was clear the Web had all the momentum.

"Even still, people were producing "My Favorite Links" and home pages, not apps."

And the buzzed-about HyperCard stacks were always more like presentations than apps. So?

I'm not trying to run down HyperCard here, it was a brilliant, revolutionary product. But I just can't fathom how anyone could think it more accessible to newbies than the Web was at the same time. And the Web has only gotten more accessible in the years since.

Your examples all explain market forces that led to the web's rise -- rather than technological reasons.

> True, but (1) PCs were significantly cheaper than Macs, and (2) many, many more people already had PCs on their desks than had Macs. So unless you were one of the few who already had a Mac, getting started with HyperCard meant plunking down money for new hardware.

In 1987, the web wasn't invented yet, and Windows 1.0 was the best the PC had to offer. This had some interesting results in the marketplace, but doesn't really explain much other than the fact that there's value in cross-platform implementations.

Apple wasn't interested in making HyperCard/Claris cross-platform, so they didn't.

> Again true, but irrelevant. We're talking about contemporary mindshare, and starting in the early '90s HyperCard and the Web were competing against each other in real time. HyperCard had a long head start and yet even by 1995 (the year Netscape 1.0 broke out) it was clear the Web had all the momentum.

I remember that time quite clearly, and until the mid-to-late 90s, HyperCard and the web were in COMPLETELY different spaces.

People were building complex enterprise database-backed applications on top of HyperCard; meanwhile, on the web, we had static pages, and post-1993, if you were lucky, CGI-based e-mail forms and visit counters.

Even with the introduction of JavaScript in 1995, we didn't see much other than rotating banners in the status bar until much later. Simply put, the web was not being used for applications at that time.

By the time the web did "break out" in 1995, Apple had moved HyperCard to their Claris subsidiary, stopped producing updates, and started charging for the authoring tools (a funny thing to do, given that HyperCard had always been Smalltalk-like in combining authoring and consumption).

Essentially, the lack of updates and unreasonable pricing led to its demise. These are market forces, not technological ones, and do nothing to disprove that 26 years ago, there were tools accessible to normal humans.

> Even still, people were producing "My Favorite Links" and home pages, not apps.

I first went online in 1991 (so that my friends and I could keep playing Shadowrun all winter) and got into the web in 1993 (I was 16, pictures were a big deal). Back then, there was this sense that the web could be really great, but I don't think many people grasped what it could be used for. Search technology wasn't great, so starting on someone's home page and surfing was how most people explored the web.

The first thing I ever built on the web was a homepage for a Shadowrun character. It was basically walls of text in which I talked about his complete history, and a whole lot of links to gaming resources. Needless to say, I got all the girls in high school....:)

Thanks for taking me down memory lane - sometimes I forget just how incredibly excited I was when I first discovered the online world. Without a hint of hyperbole, I wouldn't be who I am (in fact, I might not even be alive) if not for the web. :)

Your first argument seems to undermine your second argument.

Being an egalitarian, I have to keep reminding myself of this, even if I recoil at how messy some of the results are. Still, it behooves us to pause and reflect, and maybe consider if there is a better way? The first comment I read in this thread said that Obj-J and Cappuccino were hell to deal with, and much as I scoff at the idea, some people (read: a whole lot of web developers) aren't willing to spend a couple of days learning a new toolset. What we need to do is take the good stuff, the optimized, polished debugged and proven software and move it beyond merely libraries, toolkits and frameworks: we have to make it easier to use. This same dichotomy exist(s|ed) between PHP and everything else (eg, Perl, Python, Ruby), and inroads are being made there; why not try to make the same inroads with our web frameworks?

I can't say that this is a problem specifically in the web.

The landscape is littered with C frameworks, C++ frameworks, Java frameworks, .NET frameworks, and so on for every other language, that serves to abstract so much from the basics that it becomes so difficult to do basic things the <insert framework here>-way. The problem is that instead of being ancillary and supporting productivity, they are opinionated and try to do EVERYTHING for you, which just makes them really crappy DSLs.

This just happens to occur on the web a lot more because what we have to work with is inconsistent, sometimes poorly implemented, and confusing as hell to new players. So everyone and their dog tries to "solve" developing web applications by making you believe the DOM doesn't exist (jQuery, MooTools, etc). Some think HTML and CSS are too complicated and try to get you to believe those don't exist either (Ext, etc). Then some want you to forget that Javascript ever existed, or to make Javascript less Javascripty (GWT, Prototype/MooTools, etc).

The end result is a lot of opinionated dogma that rarely actually help us get things done (or get in the way more than help). If you pick a big framework, it's usually very good at whatever it was designed to build. But if you want to build something even slightly outside of that set, it becomes a massive cluster-fuck of WTF-inducing horrifying debugging and pain. Only the lightweight libraries that are massive value adds without enforcing a specific dogma on the user have been extremely successful.

The problem is that everyone is too opinionated and makes their software try to do your job for you. It's not really about the ease of use, it's about the ease of integration, extension, and whether they make our jobs easier or nightmarish. Yes, the web foundation is far from perfect, or even great, but most of these frameworks are far, far worse, since they massively and arbitrarily limit what you can do for the sake of some vision of simplicity.

The same problem exists in the native world. There is no consistent presentation layer that rivals the web, and even it isn't fully consistent (all browsers are different, and even the same browser on different platforms can behave differently). But it's a far cry from anything else.

As an aside, since JS isn't assembly, using it as a compile target to build applications foregoes a lot of the other fundamental technologies in a web browser. Sure, you might be able to compile <insert language here> into JS and have an application run in the browser, but that necessarily implies you're doing things like building the DOM manually, and sites like this tend to have significant performance issues. If you want it to work well, the best way currently is by using HTML/CSS/JS and by not letting a gigantic framework pretend that those things don't exist.

> Thirty years ago these people would have had no options other than to be passive consumers of software other people made

Really? Thirty years ago many of the most popular home computers booted directly into a BASIC interpreter.

Programming hasn't become more or less accessible with the web. You just write JavaScript instead of BASIC nowadays.

It is as if with the birth of the web there was a complete reboot of the IT industry

The web won. It fought a battle against many foes, and it came out on top.

Re-read that again and seriously think about it, because comments such as yours don't accept that simple reality. Do you contest it? Do you argue that the web somehow circumvented the competition?

For 15 years I've been promoting the web with groups that I've consulted with or led. For 15 years competitors have arisen with various alternative strategies, including MFC applications, Java Applets and then Web Start solution, .NET Winforms and then .NET WPF applications, Flex and Flash applications, Silverlight applications, and on and on.

And they all fell by the wayside. They lacked in features. They lacked in agility.

The only competitor that has made serious inroads are native apps on mobile devices, eking out an advantage due to the unique profile and unique input techniques of those devices, and of course gaming given that the web had no interest in gaming. The web is adapting.

If the advantages that you hold are true (in another post you argue that we salivate over stuff that was trivial 20 years ago. Yet ignore that we do things on the web that are just incredibly rich with but a few lines of markup and some ancillary code), those apps should have long devastated the web, teams making secure, easy to construct apps that run circles over those poor web developers.

I feel the true 'reboot' and memory loss happened with native mobile apps. How did we ever think this would be a good idea? As you mentioned, the unique inputs seemed to have temporarily blinded us to real progress we've made with the web. Why can't web-based apps accept all these inputs as well?

The other big problem we have is that we're stuck with this idea of having cute little icons to launch things. Why can't we have icons that just launch a specific URL? Because then App Stores wouldn't make any money then would they?

Edit - Would someone care to explain their downvotes? Surely HN users wouldn't think to use a downvote to express that they just disagree, right?

> Why can't we have icons that just launch a specific URL? Because then App Stores wouldn't make any money then would they?

That's what iOS 1.0 did. There were no native apps, just web apps that had cute icons that launched a specific URL.

People tried very hard to make that work. It didn't. The breadth and depth of mobile applications we see today only occurred once Apple provided a native SDK.

The reasons for this are simply technical ones.

I think Palm & WebOS were on the right track by making their native apps still based on web technnologies. As far as I know there weren't any mobile inputs that those apps couldn't handle. Blaming limitations on tech seems silly. The decision was a business one.

Besides that, why take away the ability to create webbased apps? Could we not have the option if I know I don't care about any of the fancy stuff. There are so many apps out there these days which are just shells of an app with a WebView inside. It feels a bit ridiculous to have to buy Apple hardware to make a shell and then use our preferred dev environment to develop the site/internals of the app.

You can just write web apps for the iPhone, even today. And yes, you can create an icon in Speongboard so that it looks just like a native app. That's how I use Google Maps on my iPhone.

You're talking about 2011 storylines.

There were plenty of discussions about HTML5 vs native mobile apps. Big companies like Google and Facebook extolled HTML5, partly because they needed leverage against Apple's stronghold over the dominate ecosystem. So yes, the decision was a business one to begin with.

Then when reality didn't fit their agenda they've all since switched positions in 2012. Facebook rebuilding their iOS app with Obj-C, for example, to be as fast as possible. So then the decision ultimately became a technical one.

Maybe it'll flip flopped around soon. We did go through a period in the 90's when CD drives came out and everyone rushed in to deliver as much multimedia content on CDs as possible before that industry died and gave way to the ubiquity of the Internet.

In the future, smart phones are the CDs.

    >I feel the true 'reboot' and memory loss happened with native mobile apps. How did we ever think this would be a good idea?

Native apps weren't a good idea for innovative reasons or to push boundaries of what's achievable, they were a good idea for business reasons: vendor and platform lock in.

And yet web apps don't seem to solve vendor lockin either; see Facebook, Twitter and most recently (and poignantly) Google Reader. And before you mention that people were able to move from Google Reader to something else, you might want to consider it was because it was based on an open protocol (RSS) that wasn't tied to either native or web.

Windows also 'won'. That does not make it the best solution. Merely the most successful.

What is your definition of 'best'?

Unfortunately, this argument (which tends to imply technical excellence should trump all other factors in the success of a technology) has been used in technology for a long time, but frankly, in the end doesn't matter.

Ask Betamax, OS/2, etc, etc.

I have very bad words against MS on the OS/2 one: http://yuhongbao.blogspot.ca/2012/12/about-ms-os2-20-fiasco-...

Windows won...then lost to the web.

> The web won. It fought a battle against many foes, and it came out on top.

It won what, exactly? The war against Gopher for presentation of hypertext documents and information display? There's more to heaven and earth than data presentation and consumption.

The examples of genuinely great web applications are few and far between, if not near non-existent. Are Google Apps really the best the web has to offer?

Cappuccino provided one example of how to move the browser past the shackles of the traditional HTML/JS/CSS and into a world where developers could work with more accurate abstractions that actually represented the problem domains of application development.

This lesson has essentially been ignored, tossing out 3+ decades of experience our industry has in building applications.

There's more to heaven and earth than data presentation and consumption.

I do my banking in a browser. I do my email in a browser. I manage photos in a browser. I manage relationships in a browser. I do my taxes in a browser. I order goods in a browser. I ship goods in a browser. I reserve items at the library in a browser. I watch movies in a browser. I listen to music in a browser. I push apps to my smartphone in a browser. I get directions in a browser. I find phone numbers in a browser. I phone people in a browser. I video conference in a browser...

I can go on and on all day. If you really want to argue that the web is a surrogate for Gopher, this conversation is futile.

> I do my banking in a browser. I do my email in a browser. I manage photos in a browser. I manage relationships in a browser. I do my taxes in a browser. I order goods in a browser. I ship goods in a browser. I reserve items at the library in a browser. I watch movies in a browser. I listen to music in a browser. I push apps to my smartphone in a browser. I get directions in a browser. I find phone numbers in a browser. I phone people in a browser. I video conference in a browser...

I'd be interested to hear how those tools stack up against their native (or cappuccino) counter-parts, because implicit in your argument is that all these attempts to extend the web into the application space are better (and have 'won') over traditional approaches to application development, which I can't say has been my experience, even for the 'best-in-breed' solutions such as Google Apps.

Personally, I've found the surge of interest in native development via mobile to be invigorating, since it has meant a resurgence in applications that put user experience ahead of dogma (or even cost). I can hope that this continues to carry over to the desktop and other spheres over the next 5 years.

You don't have to install the browser based ones. Also, if they are decent at all, you can link to them.

You can push apps to your phone with a native app? Maybe if you're using iOS. You can reserve items at a library with a native app? Which one? I don't know of any; do you? What do you use to manage photos, iPhoto? Is it really that much more powerful than flickr -- or even Facebook?

For many web-based tools, there aren't cappuccino/desktop analogs. They just don't make sense.

The success of the web makes horrifyingly apparent how important ubiquity and ease of deployment are. In they quest for users, they trump almost everything else.

It's awesome for things involving communication and relationships (banking, purchasing, and so forth) and for accessing content (Youtube is my main music discovery service, for example).

But I don't like doing everything in a browser. All the things you mention are transactional, but when I'm (say) writing music I much prefer to be in a native environment because the transactional stuff is a much lower priority. It's distracting, for many tasks. At the very least, I want to run some things without a browser frame of tabs, URL bar and so on without necessarily doing so in full screen mode.

> At the very least, I want to run some things without a browser frame of tabs, URL bar and so on without necessarily doing so in full screen mode.

Have you ever tried a Chrome Platform App? They don't have any browser chrome--they're just regular app windows, painted with whatever the developer likes--but they're still running in a browser. Sort of similar to Air/Silverlight.

>> I do my banking in a browser. I do my email in a browser. I manage photos in a browser. I manage relationships in a browser. I do my taxes in a browser. I order goods in a browser. I ship goods in a browser. I reserve items at the library in a browser. I watch movies in a browser. I listen to music in a browser. I push apps to my smartphone in a browser. I get directions in a browser. I find phone numbers in a browser. I phone people in a browser. I video conference in a browser...

I do many of these things in a browser as well, not because the experience is so great, in fact, most of the time it's pretty horrible. Banking, photo management or video conferencing in the browser a great experience, really?

I sympathise with this article, and even though web applications have come a long way, I'm sometimes amazed by the collective cognitive dissonance that seems to be prevalent among people who develop web applications, pretending that all is fine and dandy in web application land, and we've long surpassed the point where web applications provide better experience than good native applications. They are simply nowhere near, and the fact that e.g. Google Mail (which is a great web application) is considered superior to e.g. something like Sparrow is quite frankly very amazing.

> the fact that e.g. Google Mail (which is a great web application) is considered superior to e.g. something like Sparrow is quite frankly very amazing

It's a good example actually. Even considering the less pretty UI, Google Mail is superior to Sparrow precisely because it is web-based. From the user's side, the way it deals with spam and search are two huge functional advantages which would be difficult to impossible in a native app. And from the vendor's side, the ability to manage updates and monetise through ads are clear benefits which keep the program running.

Why are 'spam and search' difficult-to-impossible in a native app?

For that matter, why are you equating 'the web' and 'server-side'? The standads IMAP protocol stores messages server-side while also enabling effecient server-side search. The standard managesieve protocol implements support for managing server-side filters from remote clients. Tools such as SpamAssassin perform intelligent per-user filtering server-side.

At that point, the only real differences between webmail and native clients are:

- UX

- Native clients use standardized protocols by default.

For one thing Spamassassin is quite hopeless compared to Gmail's filter (and yes I have compared). But more to the point I would absolutely equate web with server-side hosted app. For example, if you want to implement decent search it's probably going to be Lucene-based. So if you are just talking thin-client presentation layer, the app isn't really native in any case - the acid test for a non-web app is whether it works without an internet connection. (Other major differences would be ubiquity and maintenance.)

I'm pretty confused. Are you saying that my mail client is non-native because it uses IMAP server-side search?

Even though it caches all e-mail locally, works offline, and it is a native app?

>> From the user's side, the way it deals with spam and search are two huge functional advantages which would be difficult to impossible in a native app.

You can configure Sparrow (or whatever native e-mail client) to simply connect to your GMail inbox and have all the spam filtering and search capabilities that GMail has. Even the default Mail.app on OS X and iOS (which isn't exactly feature rich otherwise) does this. As much as I like GMail as a service, their web front-end is actually the part I dislike most about it, which is why I almost never use it.

>> And from the vendor's side, the ability to manage updates and monetise through ads are clear benefits which keep the program running.

Well, first of all I don't like ads and I don't care about how service providers can monetize it because it's web-based. I'd much rather pay a monthly fee for the service and/or a one-time price for the app, than having a web UI full of crap I didn't ask for. Not a very strong positive for web-apps. As far as updates are concerned, I understand that a web-based service is much more flexible in that regard, but then again, it's not really hard to have self-updating native applications either.

I do my interactions with my bank in a browser, but my bank doesn't do the actual banking in a browser.

I could manage my photos in a browser, but I won't do "photoshopping" in a browser.

I watch movies in a browser, but I don't edit my home movies in a browser.

I listen to music in a browser, but I don't record or mix music in a browser.

I push apps to my smartphone in a browser, but I don't develop smartphone apps in a browser.

In general, I communicate or consume in a browser, but the worthy things - the creation - tends to happen outside of it.

Also, don't forget that all the things that consumers might ever see are just the tip of the iceberg - the vast majority of software and hardware development is done not for B2C apps/websites but for solutions that are B2B, business-process-internal or hidden in an appliance, never seen by any consumers.

I think you will end up doing every one of those things in a browser.

That browser will probably look less like a browser - since it does everything, it doesn't really need a window.

*Except for your bank doing your banking. Of course it will be offloading the work to a dedicated application. The rest, however, will be browser-based.

Enterprise B2B has started moving very strongly towards web-based apps in the last couple of years e.g. Salesforce

Actually I do most of that stuff in native mobile application now instead of in a browser.

Do you know why? Because the browser version sucks ass compared to the native version someone has designed for a phone UI. It makes me sick every time I pick up my phone to transfer money in my bank account while I sitting at a computer.

It's like there's a culture of cultivated ignorance around web apps.

The pop culture of web development strikes me as a spastic, young programmer who flits from framework to framework, always in search of some magic elixir to make application development more pleasant. They know it isn't quite right, but they don't want to give up the hard-earned, arcane JS knowledge they've used. Nor are they willing to sit down and admit that, maybe they need to sit down and learn a thing or two about how to structure programs better (as we all do)!

Instead, it's emotional blog posts and bikeshedding over inane topics (vim, Coffeescript), rather than looking at the deeper issue - the utter mediocrity of the tools they're using.

I've been doing software development very successfully for closing on two decades. I've developed in 680x0 assembly, x86 assembly, C, C++, MFC, ATL, Java, .NET and on and on in the fat-client space. I've watched with interest and anticipation as various parties have brought out the Next Big Thing.

I am also a strong advocate of web apps and have been for a long time. Reality has proven out my position as, as has mentioned, the web won the presentation tier (clearly this has nothing at all to do with data processing or database or system programming or anything of the sort).

In any discussion on web apps they are always held against an illusory foe that is all capable and all perfect. Let's get specific however -- I love Rdio and love their web app. They came out with a new .NET client so I of course gave it a whirl.

Worse, slower interface. Constantly updating (in big monolithic chunks). Consuming hundreds of MBs of memory.

I'm not trying to criticize specifically Rdio, but I guarantee that initiative started with someone slamming the web and its crazy mishmash of standards, promising that they'd do something that would exploit the full power of the client. The result was something much worse, holding the single and only advantage of "not in a web browser" (for what that's worth?). Amazon has fielded various trials with various rich apps (they usually partner with Microsoft), but you've never seen them because they never have a compelling story, coming with enough detriments that it outweighs the benefits. My local library system used to have a whole nasty Java interfacing tool that was an absolutely abomination, with a horrendous rickety interface that seldom worked properly, was slow, and was a beast. They replaced it with a dramatically better web interface.

Those are all just anecdotes, but they're constant across the space. Yet every post pretends that the web is a weakling that has no merit and no advantages, completely missing why it has made all of these wins against endless competitors.

> inane topics (vim, Coffeescript)

Sorry, you're going to need to elaborate here or else you're really sensationalist. Also, you can't be serious about the community not admitting we need to learn how to structure programs better, as if that were the only problem. I didn't learn that UNTIL I started working on the web.

horrible but open , and i think that matters. After the flash and silverlight fiasco , though these tools are far more powerfull , you just cant trust a private business to provide you these tools on the long run. My alternative to that mess is Haxe, and it's going to be big.

> The examples of genuinely great web applications are few and far between, if not near non-existent. Are Google Apps really the best the web has to offer?

Every good web application beats its desktop rivals simply by being ubiquitious. People who dismiss the web typically ignore this simple fact completely. But it's always been true, you can look back to 1997 when Yahoo Mail killed the desktop email client business not by being flashier or better (it had full-page refreshes over 56k modems, awful!) but because you didn't have to ever worry about not having access to your email again.

I agree. Here are two (old) essays that really sum up this argument for me:

How Microsoft lost the API war: http://www.joelonsoftware.com/articles/APIWar.html

and

The Location Field Is the New Command Line: http://daringfireball.net/2004/06/location_field

Note that both of these articles kvetch about user experience and how the web simply can't compare to native apps, but emphasize that the ease of deployment matters more.

Also, note that the app store model obviates some of the deployment difficulties of native apps and makes it almost as easy to install native apps on the phone as it is to visit a web page.

(Apologies for the big rant, these are some ideas that have been rattling about in my head recently and your post made some things clearer in my mind. It's not meant to be a post of factual 'truths', but an explanation for my (and perhaps others) reaction against the web as the one platform to rule them all. So it's as much an emotional gut reaction as a logical one. )

I think this (desire for ubiquity/multi-platform) is the main reason I don't see eye-to-eye with the people who are pushing for web apps for everything. Web apps everywhere is great for large corporations that want to provide their service on as many platforms as possible. It's great for venture capital funded startups where the name of the game is to get as many users as possible and get bought out before your money runs out.

I'm a small indie dev though, I produce software (not software as a front end to a non-software business), and I'm not looking to become a billionaire (really!). I want to develop software because I enjoy it: to solve problems for real people, to solve my own problems. So really I don't care too much about being able to target every platform at once. I want to target the platforms I like, using the tools that I like, and as long as the platform has enough users to make it viable then I can do that with native software on open platforms (the latter is important and why I don't write iOS software, for example). There is the risk that the platform I target gets killed, but if a lot of people use it then that's generally going to be a slow process, and there is plenty of churn in web tech and standards anyway so it's not like it provides immunity to technological change. Additionally, I just have a hard time working with web technologies. I find them much harder to use, to the extent that if using html/css was the only way to write software I wouldn't be writing software at all and I'd go stack shelves in the supermarket instead. Maybe I'm just not smart enough to use them effectively, I don't know.

I guess this is partly the old 'software as craft'/'lifestyle business' argument, but it saddens me slightly that the big industry players are moving us away from the kind of industry where you can write software that interacts with some arbitrary bit of hardware in unsupported or unintended ways towards a much more controlled and profit orientated one (i.e. your software running everywhere being the most important criterion of all). Web people are not the only offenders here of course, but arguments against things like iOS have been well explored over the years.

Some people seem to view the web as some kind of beacon of freedom, the little guy fighting the big bad industry giants of Microsoft, Apple, Adobe and so on. But from where I am, the web is the big bad industry giants. Writing web software means being restricted to a very limited sandbox. The APIs available to me are decided by what a small cabal of powerful companies (Apple, Microsoft, Google, Mozilla, Nokia and maybe a few others) can agree on. None of their interests align with mine (why would they?).

If Gopher had taken off and supplanted native development to the extent that people were using platforms that could only run Gopher apps, would the current web even exist? All user facing software being written to a single standard and API has good things about it, but it also limits new ideas tremendously ("sure you can add X feature you just need to get it through this committee of industry heavyweights first and get them all to agree, and then wait for them to ship it").

What is important to me is that data uses open standards, whether at rest or in motion. If people control their data, and it is in open formats, then there is little companies can do to lock them in. Which is good for me as a dev since I can write software to help users work with that data. Web technologies don't do much to help me there, in-fact for the most part they have made it easier for companies to keep users data locked away on their server in unknown formats or behind proprietary APIs that I don't feel confident building software against. More concisely: email good, Facebook bad; FLAC good, Spotify bad.

I'm not arguing against all web sites/apps (or any hypothetical tech stack that offers a similar standard sandboxed environment). For some things it can be appropriate: wikipedia or hacker news or other inherently networked activities for example. I'd prefer something less byzantine of course, but I expect that if the web apps model does take off then ways of targeting it that mean you don't have to think about CSS or Javascript will get more traction, in the same way that I can write native code without having to worry about assembly (most of the time). Javascript as a compilation target will become "fast enough" on most hardware for general purpose things (not for absolutely everything, but that is just another manifestation of the problem of not being able to break outside the sandbox when needed). So the technical issues of css/javascript being horrible are not the main problem.

I come from a social science background, which perhaps taints my view here. I look for the power dynamics that limit my real world ability to do things, and don't really care so much about theoretical ideas about a perfect world where everybody works together in harmony to make the web a platform that does everything you could ever need in a timely way. My issue is about power: who controls the APIs that ultimately lets software control your hardware and data? With open native platforms the answer to that is anybody. With the web it's the tech giants alone.

Interesting perspective and good food for thought.

My one point would be perhaps the way things are now is because it's good enough. Using the web and building for it is good enough for a lot of people, the speed bumps and systems in place are good enough for the majority.

But I do agree, we've gotten to a point where a handful of corporations dictate how everyone else get to do the neutron dance and that the bleeding edge needs to be examined and worked on so that change can occur.

This is interesting enough you should write it up as a blog post and submit it. You're not going to get the discussion you deserve posting something this interesting deep in a thread on a tangentially related article.

Most desktop applications that where written for the desktop back in the 90's and early 00's where boring internal business stuff, not well known shrink-wrapped stuff such as MS Office.

Today, almost all that development has moved to the web.

It's almost as if you've drunk your own kool-aid.

>And they all fell by the wayside. They lacked in features. They lacked in agility.

No, no and no. The web "won" because its simple and the lowest common denominator was good enough for most people. The web has always been far behind in terms of features and agility (depending on what you're trying to optimize when it comes to agility). The web won solely because it was "open" and every company had a stake in ensuring no one company could sustain their monopoly. The web sucks and it did not win on its merits.

But it has won nevertheless. The question then is what we are to do to move the web into the realm of applications. I don't have an answer, but one advantage mobile apps hold is UI standardization.

There is no one right way to build an app for the web the way there is for Android or iOS. That's a problem for both the applications authors and users. Cappuccino attempted to solve that problem in a way that did not appeal to programmers, it seems. jQuery UI succeeded to a point, and now Bootstrap seems to be the framework of choice. Perhaps a widget-based UI framework that can target the desktop and mobile devices would make sense.

Being ubiquitous is a merit.

Which is one of the strongest selling points of Emacs and Vim. Ubiquitous doesn't equate to web based; there are plenty of native apps that are ubiquitous, and on top of that they don't require a constant network connection, or that the server is up. It also seems to me that if you have the source, you can get the software running just about anywhere.

I don't think we disagree. It's just that on the web ubiquity is free.

It just happened to be that all browsers adopted HTML/CSS/JS and now we use that as an application platform (despite it being designed for sharing documents). It won because there is no other choice but it kind of still sucks which is why there is so much complaining about it.

>The web won. It fought a battle against many foes, and it came out on top.

What utter BS. You are essentially celebrating the victory of consumer commerce, not technology X over technology Y.

Think about it.

There never was any battle between the "Web" and "MFC, Java Applets, Webstart, .NET, Flash, Silverlight". Those thing survive & even thrive in places where nothing else would do. Do you really think everybody in investment banks & brokerages fire up their browser first thing in the morning ? 99% of what they did, do & will continue to do, will continue to be done via native apps - mostly C#, C++, .NET & Java apps. This is just plain reality. It isn't because they are lazy or don't grok the web. Processing realtime trading feeds isn't the browser's strength. Sure I'll show you an iframe with some ticker, equities blotter, trading tickets ...but that's just consumer commerce. I'm soliciting your cash - Here consumer, click on this Buy button, give me your money. After that ? Once you fire off your client request, that's when 99% of the task actually begins, and it ain't on the browser, no siree.

The web is just the tip of the iceberg. Most of the action is below the surface of the water, and that's not the web & will never be.

To your point - the whole idea behind Webstart, Java applets, Winforms, was to mimic the native UI on the web & try to offload as much client-side logic to the client's machine.

Turned out customers didn't like "fat clients" - they wanted textfields & simple input forms, accept their inputs & go away & process the inputs elsewhere.

We said - Hey, you guys have a fast processor & slow network, let's do this one-time applet download/Webstart download/Silverlight app download, & once done, we won't tax your network, we'll just tax your machine. I wrote one of the first equities trading applet for Goldman Sachs. It was a fat client ~100 MB, you actually waited 10 minutes for the install. But it did so much - all the trading rules validation, portfolio construction & computation, everything happened inside that applet. Once the applet downloaded, it felt & behaved like a native desktop app. It was so much more powerful. There was actual science in that applet. It took a doen CS PhDs to build the infrastructure for shipping that. We fucking got dozens of patents...pseudo-realtime trading applet over the web ha ha ha! Today, the interface is so dumb - zero download time - but they hit the network so much harder. Absolutely nothing happens on the client. To check whether a simple trade is valid, you have to hit the server multiple times during the workflow. Customers are ok with it because the network has gotten so much faster. They were downloading ~100MB applets on 56.6k modems, now they have superfast broadband, so it makes sense to hit the server more & do less on the client box. So you just hire an 18 year old & he'll build a polished web UI for the trading applet over a weekend. That doesn't mean all those CS PhDs just up & died. They just migrated to the server side, where all the action now happens.

So this isn't success of the web & failure of the applet - its success of broadband. Network penetration today is massive & ubiquitous. You can place million dollar trades over android. You couldn't do this just 10 years back. At that time, applets were state of the art. In hindsight, the whole exercise appears pointless. 100MB applets WTF, just use underscore dot js & ajax & broadband & a little bit of markup & your trading window shows up on a tiny 4 inch android cellphone. Yeah, true, but we didn't have any of that back then. Cut us a little slack, will ya.

"You are essentially celebrating the victory of consumer commerce, not technology X over technology Y."

It isn't celebrating to point out a reality of the world relative to a post that seems to imagine that every other solution was handicapped in the race. And it IS completely based upon technical merit.

"Do you really think everybody in investment banks & brokerages fire up their browser first thing in the morning ? 99% of what they did, do & will continue to do, will continue to be done via native apps - mostly C#, C++, .NET & Java apps."

Humorously I build software for the financial market. Our presentation tiers are almost entirely through web apps. The competitors I was talking about -- the ones who poured cement on their own feet -- went the "thick client" route. And we keep on iterating.

"The web is just the tip of the iceberg."

Yes. No kidding. I'm not building a database in the DOM. I'm not processing trades in JavaScript. This discussion is about the presentation tier. Every criticism of the web always holds that you could toss together something so much better in some illusory, never proven alternative, but that simply isn't true.

Yup, I build software for the financial market too. Years ago we were struggling to persuade companies that they should use html instead of sliverlight or flash. Last year, they pretty much all realised at once that they'd built millions of dollars worth of key software in dead technologies (and often struggled too). They're all back now, wanting products and help with their HTML/JS.

I love how you turn this into a talk about re-solving old problems to how mainframes are the way of the future.

Kind of sounds stupid without 6 paragraphs filling it out, huh?

I broadly agree, but I think perhaps your thesis is a bit premature given the latest push for mobile apps, and (perhaps more dangerous to the web), walled gardens like Facebook.

But I think the open web will win in the end, not because of any inherent technical advantage (in some ways it's behind native apps still in terms of performance, look and feel etc), but because the open web has so many advantages for developers and users that in the end it will overcome its disadvantages. Performance and UIS are improving constantly on the web (contrary to the opinion of this poster), so the gap is narrowing. At some point the performance won't matter, offline web work will become possible, the network will be fast and ubiquitous enough and the advantages of the open web will come to the fore.

As a developer of web apps and mobile apps, the contrast between deploying a few fixes a day to a website, and working hard on new app versions only to have them rejected by an anonymous gatekeeper at Apple, often for inscrutable and unjustified reasons, then waiting days or weeks for approval, could not be more marked. Things are easier on the desktop, but there are still far more hurdles to deployment, and that matters a lot, as you hint above. The only thing I really miss on the web is a great language to script web pages in, but that will come (js is not my favourite language), and you can use any language you want on the backend, and no-one need know.

As a user, the advantages of the open web (i.e. not walled gardens like Facebook) will I think have an irresistible pull - advantages like linkable, shareable content, content on any device, even unforeseen ones, not having to rebuy apps but just subscribing once and accessing content from anywhere. The fact that the web is not controlled by a single vendor as most commercial operating systems or devices are is an added bonus.

Finally, I'd argue that the radical simplicity of web technology is actually its greatest strength, as it means it can be deployed almost anywhere, even with limited resources, and give at least some value to users for decades and over any new devices which come out, as opposed to the binaries being created for current platforms which in a decade will be forgotten or dumped by their vendor in favour of the newest flavour of opendoc, .Net, whatever... I find it fascinating that the web delivers consistently better results by virtue of keeping the tech simple and backwards compatible.

Worse is better after all.

>That's spot on

I agree that web developers in general and web development in general follows this trend. But his evidence for it is unconvincing to non-existent. Caring about file size is not a symptom of "tossing out everything we've learned from the last 30 years".

I disagree. Web programming can be fundamentally different from application programming.

Application programming tends to be far more complex, far more planned, and far slower. Applications last for years and years, and new releases might take multiple years.

Whereas web programming can be extremely experimental, prototype-y, with multiple releases a day. It's also vastly faster to program.

This isn't an argument against good architecture -- far from it. But it does mean that the kind of long-term planning and investment that makes good traditional applications, isn't always applicable to website programming.

Due to the nature of the market, "complete disregard for maintainability or extendability" is sometimes a good thing, because you can iterate faster. Think of a lot of web programming as prototype programming -- then, you go back and re-architect, re-engineer, etc., as desired.

"Application programming tends to be far more complex, far more planned, and far slower. Applications last for years and years, and new releases might take multiple years."

Really? You can build non-web apps just as fast as web apps. There are decades of tools and programmer experience doing this daily. You can't build Autocad or Photoshop as quick as you can churn out a buggy Rails site, but that's comparing apples to oranges.

> You can build non-web apps just as fast as web apps.

I expect the underlying point was that most popular native platforms today almost force you into best practices which does add some upfront cost.

On the web you can throw up a single page PHP script full of SQL injection vulnerabilities and Javascript spaghetti and you are off the races. There is generally no equivalent for a modern GUI system. You typically need to start with a full MVC infrastructure in order to even begin to interface with the system libraries.

When you take care to craft your website using the same best practices the differences become negligible, but a large number of developers do not care about architecture and can take shortcuts that make it feel like they are moving faster.

"comparing apples to oranges" that is kind of the whole point here.

Most of the arguments in this thread are coming from two totally different perspectives. Some holding doggedly to various ideals and dogmas and instead of trying to learn the new domain, they try and make the apples look like oranges. This never works. Most are right about this or that when it comes to the domain that they are experienced in, but when economics forces people into a new problem space some don't try and learn the new space as a thing in and of itself, instead they try and adapt their methodologies and abstract to a different problem space they are more familiar with. That is where these conflicts come from.

The believe expressed in first three paragraphs proves

> all the lessons of the process of developing software have also been discarded and are slowly being re-learned.

"Cool URLs don't change." Quickly fixing my server is great and all, but the fact that I can push a button and instantly break millions of hyperlinks is a fundamental problem with the architecture of the web.

It's not that there was a complete reboot of IT. It's that web solutions are just now being applied to problems we already solved years ago in other industries (telco, defense).

The web wasn't originally intended to be used to provide native like experiences and so the foundations for that just weren't there. So yes, people had to re-invent and re-discover all those desktop frameworks but this time in a javascript & html environment.

I think the obsession over desktop like frameworks in Javascript is due to the fact that there just isn't a clear winner yet and that is because the foundations required to support a clear winner just don't exist yet.

-We need the ability to scope CSS to a given section of the document. I want to be able to import someone's widget (or set of widgets) and not worry about style collisions.

-We need a better solution for storing templates (pre-compiling them into .js files or storing them in <script type="text/template"> just seems awkward)

-We need a better way to bundle CSS with its resources. I.e., if I have a css file in "components/bootstrap/styles" that refers to "../img/pic.png" it is going to break when I concatenate all my CSS files into one and drop them in "/"

-We need a better module system. Yes, requirejs is good but I'd like to say I depend on "package.name.x" and not care where it comes from or who provides it. It should just get loaded for me (this has been solved for years in the Java world via OSGi).

-We need a better build system. Yes, grunt is good but I have to put in all sorts of logic to handle the above problems.

-I should be able to place custom tags in HTML and have the browser create my user-defined component for me. I.e., <usr:combo> should find the correct template and backing javascript and construct the component.

There won't be a winner and the re-invention will continue until these foundations are put in place.

Good news! Many of these very problems are being worked on right now. The component issue you raised is being fixed with web components: http://www.html5rocks.com/en/tutorials/webcomponents/shadowd... which should be making its way into stable browsers in a matter of months. The module system is being fixed by ES6 modules which should be appearing in Firefox and Chrome within the year. I haven't heard anything about namespaced CSS and I agree that's something that we need.

Even as someone relatively new to development and a native of the web this seems so clear to me. The web was originally just a way of sharing documents. It is both a miracle and heresy that we have managed to do so much with it. We could very well do with some foundational rethinking of what the web is beyond all the web 2.0 buzzwords.

That said, I have never understood the driving need for desktop like frameworks. The problems that ubiquitusly networked applications and communications technology face are nothing like those of kernel code, commandline applications, or modern applications like CAD Photoshop. All the things you mention that need fixing desperately need it, but I don't think a desktop like framework/library/standard for the web will ever 'win' because the web is not trying to solve the problem of a virtual desktop, it is trying to solve the problem of a non-homogeneous ubiquitously networked data exchange layer.

Personally, batteries-included frameworks like Cappuccino scare me off for a few reasons:

1. I have to learn Objective-J. I've been web programming for years and I've become really good at Javascript. I even enjoy it. Other frameworks don't require learning a new language.

2. Time spent learning Cappuccino is only good for programming Cappuccino apps. Time spent in micro frameworks generally translates to other frameworks. Backbone is so bare-metal that it is very easy for Javascript programmer to feel comfortable.

3. Batteries-included frameworks often have "their" way of doing things that are not extensible. This can lead to awkward situations where you need behavior that is just not possible without monkey-patching the hell out of the frameworks and getting yourself stuck on a particular version.

4. I can't speak to Cappuccino, but a similar framework, ExtJS, has (for 4 major versions and nearly 10 years) lacked a build and minification system that didn't suck. This means you must include all dependencies, leading to 1.2MB+(minified!) blobs of JS that can take upwards of 10 seconds or more to load even on fast connections. For business-class apps running on B2B contracts, that might be fine. For public-facing webapps, this is murder.

5. People like micro frameworks because they can use every brand new & shiny piece of JS that shows up on the web without worrying about compatibility. And they can use {HTML5 feature} without worrying about compatibility. I'm sure Cappuccino does a better job than most when it comes to this sort of thing, but what happens when the devs lose interest? Years of "Objective-J" experience go to waste.

> I have to learn Objective-J.

Objective-J is a Javascript superset, so everything you know about Javascript is still available to you. It will take minutes to learn the additions. The language is not really the problem.

Its the framework that will take quite some time to learn. There are a ton of interfaces to work with. I suspect that is why you feel more comfortable with something like Backbone, because it basically brings nothing to the table, so there is very little to learn.

> Time spent learning Cappuccino is only good for programming Cappuccino apps.

Not exactly. The API is nearly identical to Cocoa on the Mac, and very similar to Cocoa Touch on iOS. If you know any of the three, you should be able to transition between the frameworks without much trouble. You can even use Apple's own Interface Builder to build your Cappuccino interfaces, to give you some idea of how similar they really are. There is also GNUStep, but that's probably not a big selling point, granted.

Perhaps the time spent is less valuable if you have no interaction with Apple devices, but I suspect a lot of people interested in web development also have an interest in mobile development where knowing how to develop for iOS does have quite a bit of value.

You're right. I'm not an Apple dev so I don't see much of the utility but I'm sure there are many who do. Certainly, if I were a native iOS programmer looking to do a rich web app, Cappuccino would be my first choice.

All of these frameworks fulfill a need (otherwise they wouldn't exist). I think the author expected that, because Cappuccino filled his needs so well, that it would fill everyone else's. He is now venting his frustration that this is not true.

My takeaway from it was not him venting about nobody using Cappuccino, it was that developers wouldn't spend even a small amount of time to evaluate it in the first place.

"if it's going to take me at least a day or two before I begin to understand, how long is it going to take the rest of my team?" (https://news.ycombinator.com/item?id=5411474)

If you have not spent that day or two to understand, how can you know it does not fill your needs?

I actually do not think Cappuccino is a good fit for a lot of the places Backbone and the like are being used. The unfortunate part comes when people will still use Backbone where Cappuccino is the right tool just because Objective-J "looks scary", or some other non-technical reason of emotion.

>Ember is very close to the “micro” side of things, but at a “whopping” 49KB (less than a second download on a dial-up internet connection

Chances are that you probably never had a dial-up connection. 49KB on dial-up generally takes around 10 seconds.

Besides this point, the accusatory tone towards front-end developers and 'the rest of the world' that did not understand you isn't helping to get your point across. You were in the wrong place in wrong time. It happens. Move on. There are better things to do, and bitterness only hurts yourself.

Perhaps someone forgot to multiply by 8.

And didn't take delays in account, half a second used to be my ping on dial-up.

I disagree. I think his post accurately gets his point(s) across. He has identified a problem and is obviously fed up, hence the accusatory tone.

perhaps that's the size of the script uncompressed, and gzip crunches it down to < 10KB? dunno for sure. could be the factor of 8 thing that another commenter suggested.

Ember is 50kb gzipped

While I agree with the main thrust of the article, in my opinion what killed Cappuccino was that it was big, weird, and rather opaque. Did it ever occur to you that maybe it was just too much? One thing I liked about jQuery was that my webpages didn't have to become applications, they could just be webpages with behavior. It worked within the paradigm, you might say, and that's what I wanted and it's what my clients wanted too. SproutCore didn't see heavy adoption either. GWT probably had more than both, but it wasn't the #1 or #2 Java framework.

I wanted to love Cappuccino, but I didn't know Cocoa or Objective-C. Your team was spread so thin I remember wondering--how much time are they putting into Objective-J, how much into Cappuccino, how much into 280 Slides? When 280 Slides shut down, it looked to me (as an outsider) as the beginning of the inevitable cascade of plug-pulling that happens when a small team realizes they bit off more than they can chew. Imagine what it would have done to Rails adoption if Basecamp had been shut down. It was the flagship product.

It's hard to imagine jQuery being shut down, but suppose it did. With Zepto and everything else, it wouldn't be the end of the world for me--my investment in it is safe. Imagine if I had written tens of thousands of lines of Objective-J. Does this blog post make me feel like I made a safe investment? We like light libraries over massive frameworks because light libraries really can't take our applications down with them. You may have brought an impressive engineering mindset to this problem, and this is a problem that deserves it, but we're still talking about 4 years from start to finish. If you want me to go "all in" on something that could take years of my work down in a ball of flame, forgive me if I'm distrustful of three kids doing a startup and trying to reinvent the world.

This was the comment I wanted to write.

A couple of years ago a London based start-up came to me asking for help finding a Cappuccino/Objective J developer. I pride myself at being able to find needles in haystacks but finding a dedicated Cappuccino dev in London was proving impossible. During a moment of desperation I fired a very simple email to Randy asking for suggestions and I got a reply within 24 hours with a few names suggested from the Cappuccino community who were UK based and a couple of suggestions on what sort of developer would find the transition to Objective J easier than others.

Randy's passion for Cappuccino was blatant and his excitement at seeing the framework expand in the UK was admirable. The fact that someone so passionate has become so frustrated is a shame.

Best of luck at Google Randy and thanks for taking time out to help a stranger.

The fact that the employer needed someone with prior Cappuccino experience is the bigger problem.

There are a huge number of frameworks out there. Pick one, and then hire people and have them spend a few days learning it.

Adding more and more frameworks leads to, well, relevant XKCD: http://xkcd.com/927/

Many of the author's complaints read similar to this one:

Now, people compare Backbone to Cappuccino, or Ember to Rails. Cappuccino, Backbone, and Rails all serve completely different purposes, and saying you don’t need one because you have another is just asinine.

Three questions:

1. Who are these people that compare those three frameworks?

2. Why do you let them get under your skin?

3. What does that have you do with you "being done with the web" (or web community)?

I don't understand any of these. So some morons want to compare apples and oranges? This has nothing to do with "the web" and crops up in all industries and spheres of life. Welcome to humanity, enjoy your stay.

As best I can tell, my lack of empathy must come from a different definition of "web community". When the author complains about how The community has tossed out everything we’ve learned from the last 30 years of building applications in favor of being able to put something together quickly with complete disregard for maintainability or extendability. part of me wants to agree with him, flip a table, and join the melee [1].

But at the same time, I was listening to the Ruby Rogues episode with Sandi Metz last night [2]. There is a 20-minute discussion about utility of does-it-all frameworks, their use in a project, the types of projects you use them for, and the point at which all projects will necessarily need to have designed a better solution, or massively adapted the framework, to better fit their problem. It's all very clear-eyed and up front about this exact tradeoff (that is, complaint by the author). That is but one example, and I won't belabor the thread with more.

At the end of the day, either the author is simply frustrated that it was hard to convince people to use his project/product, or we must live in very different "web communities".

[1] http://www.youtube.com/watch?feature=player_embedded&v=W...

[2] http://rubyrogues.com/087-rr-book-clubpractical-object-orien...

> The hilarious obsession with file size is the start of my frustrations with the web community.

It's not so hilarious if you're the one paying for bandwidth when hundreds of thousands of users are accessing your SSL site each day. Or if you're desperately trying to pull one page up quickly while on your smart phone in a region with sketchy (at best) mobile reception.

I'm one of those people who are obsessed with keeping file sizes to a minimum and, quite frankly, I find it "hilarious" when developers create sites with both a heavy bandwidth footprint and heavy CPU footprint rendering the bloody thing just because they insist users should experience a full desktop experience inside a web page. Those are different paradigms with different limitations and bottlenecks. Trying to shoehorn in a desktop paradigm (regardless of how elegant the OP might have considered Cappuccino's code) for the web simply didn't make sense back then; and that is why his project failed.

I also want to stress that I'm not against interactive sites. Obviously there is a happy medium between minimal design and the OP's vision of the web. However there have been studies that have proven that even a fraction of a second longer page load time can drive business away from your site. And given this and my former points, it makes considerable more sense to use more compact libraries to begin with.

Lastly, his point about many developers being unwilling to spend a little time learning web frameworks is something that I haven't personally encountered. But I can agree with his frustration on that specific remark if it is true.

"What do you mean the page takes a long time to render? Look at one of my monitors, it comes up in less than a second! Yes, everyone has a 6 fucking-hertz video card with 8 gigasnots of memory these days. Well, all my friends do."

Yes, the relationship between page load speed and bounce rate is well proven: http://blog.kissmetrics.com/loading-time/?wide=1

So here is my humble opinion: I discovered Capuccino two years ago. My first reaction was "This is amazing!". I need to learn another language, framework and stuff. But I thought that's the price I pay. There were three things that pulled me off, though.

1. It's buggy (or was buggy). I don't remember the details, but I couldn't setup few things because of bugs and the community was not there to help.

2. It's SLOW. Unless things changed now, I tried a few examples and I can't see how it can work. This was a huge turn-off.

3. I'll massively depend on it. jQuery is broke? Replace it with another library. This xxx.js library sucks? We can use another one. When using Capuccino you are locked to their framework. Is there a bug pulling you off from shipping? You are now screwed.

So my humble opinion is that Capuccino was not successful, at least by the time I tried it. The founder leaving the project is another reason why no one should have used it.

It was number 2 that just turned me off right now. Went to look at some of their demo apps and was met with a loading screen for 10 seconds before anything happened. Not sure about everyone else, but I know my users are not going to stick around long enough to even see the app.

I have a slow machine for now (2009 MBP), and Cappuccino is dog slow on it. I'd try it again in a heartbeat once I get a faster machine, but for now, it's literally (and I'm using that word right) unusable. A 30+ second delay after I save a simple interface in Xcode and the changes compiling. It's just frustrating. But if the turnaround time was less than a second, I think I would love to use something as "well-thought-out" as Cappuccino.

This has to be one of the most bitter and condescending posts ever. Rather than sucking up the fact that a project he has been putting his heart and soul for some reason isn't gaining traction (and trying to adjust so that it can), he blames "the web community" for all his issues. The community doesn't have a centralized decision making system, instead, people support things that make sense. In this case, they didn't support Cappuccino, because it didn't make sense to them. Rather than blaming themselves though, they blame the masses. The conclusion "screw you all, I'm joining Google" is really abrasive and arrogant. It would have been much better if this post was phrased as: for some reason our platform didn't work, here's what we learned.

As one of the people who worked on GWT, and have used other more structured frameworks (Dart, Closure), I have a strong affinity for systems with good tooling support, a consistent package management and module system, etc

However, I think what's missing is a recognition that there are different kinds of web apps. There are "application" oriented apps, primarily single page, large, and not content driven (gmail, docs, games, etc), and then there are page-oriented apps, like forums, content management systems, blogs, database backed forms/reports, etc.

The former lend them selves really well to the structured approach. You trade off simplicity and iteration speed for more of a "on rails" experience, but it helps for larger teams to collaborate on larger projects to have some restrictions on freedom and some mutually agreed upon consistent behaviors.

However, the latter page-oriented apps fit in a lot better with these microframeworks. And page oriented apps fit in with the original vision of the Web, they are documents, they have URL addressable, RESTful, hopefully, durable content, indexable by search engines, and linkable by anyone.

I think there is room for both and a sweet spot for different approaches. I often use several languages for specially suited purposes, like sed/awk/Perl for text processing, JS for small page oriented stuff or mocks, GWT for larger apps. I would not advocate GWT for people for building progressively enhanced page-oriented apps, even though I'm the author of GwtQuery and AngularGWT.

Using the right tool for the job often depends on context and personal preference. A lone hacker or small team can get by with doing a large app in hand coded JS with hand rolled build tools and module systems, but frustrate other people coming onto the project.

There is too much, IMHO, "one above all" thinking when it comes to development, and often, the original goals -- what the app is supposed to do for the end user, get lost in the zeal to build a cool framework or infrastructure for it.

  This reaction to Ember just baffles me. Your day job is to 
  build a piece of web software and you can’t take a few 
  days to learn the ins and outs?

What compelling reason can you give me to take a few days to learn Ember and not insert one of a dozen alternatives? Or should I take a few days for each of the dozens of newfangled alternatives that pop up every year? Are a few days even enough when you want to truly experience whether some framework is worth it?

Get off my lawn.

The biggest problem with Web Development today, IMHO, and the biggest reason why people desire Widget frameworks in the first place, is that the modern browser rendering pipeline is too complex for most people to understand.

If you want to create an app which can consistently animate or scroll at 60fps, avoid jank, and random hiccups, and do cross-platform between WebKit, Firefox, and IE9/10, you will find it enormously painful. And often, a solution for one particular UI design won't compose cleanly with others.

How many developers use the Chrome DevTools timeline, and minimize the time spend computing layout and the amount of repaints? How many of them arrange for optimal use of the Compositor?

This is one of the challenges in Web developer vs iOS or Android, where the UI performance is more predictable with fewer surprises. Hence, the desire for a Web framework with Widgets that is "idiot proof" when it comes to high performance. This rubs some people the wrong way because you lose flexibility to control the look and feel in many cases and must settle for building with out-of-the-box stuff, but the productivity benefits are undeniable.

It’s a huge mistake to consider Cappuccino a “widget framework”. Cappuccino’s main point is providing the whole “web development stack”. You don’t have to care about browser differences, about DOM trees, about HTML. You just write against a sane, modern, high-level API, just as you would on a modern desktop platform. The number of things to keep in your head dramatically decreases. That’s the point.

I understand how Cappuccino works (I am one of the GWT authors, and I often have to tell people the same about GWT, is it not a "widget framework"), but what I'm saying is, from what I observe, people often spend more time debugging CSS and HTML than they do code.

There are lots of reasons why JS needs fixing, I'm onboard with that. But honestly, you could fix it 100%, and it would still be easier to develop native apps on iOS/Android. 280 North's Slides for example as a great example of what you could do, but let's not kid ourselves, it was also pretty slow.

If you look at the herculean efforts the Google Docs team went through to get that running fast, and it's still not fast enough, I'd say there's there a lot more work do be done on the Web-as-app-UI-renderer model side.

If it weren't for all of the other things you'd have to reinvent, and what you'd lose from indexability and transparency, I'd almost say built a framework on using WebGL rendering.

It's a shame that 280 Slides is no longer online, because I'd love to be able to challenge your performance assertion.

Back when we released it, in 2008, browsers were a hell of a lot slower than they are now. And even then, 280 Slides launched faster than PowerPoint or Keynote on our machines, and had comparable runtime performance with manipulating objects on the slide. Of course, we did not have many of the most resource intensive Keynote features like 3D animations, but I don't think that's what you meant.

> I'd almost say built a framework on using WebGL rendering.

I've had that thought many times. But then I start to wonder why bother with the web browser at all? It begins to look like a pointless abstraction. In many ways, it already is.

I think different people have different concepts of what smooth and buttery feels like. Native apps have taught people to expect a mostly constant fixed 30fps or 60fps. As smooth as you think Slides was, I doubt it achieved this, because the tooling neccessary to monitor repaint regions and framerates, not to mention GPU compositing, wasn't available in 2008.

DOM programming is still a minefield. GPU compositing tricks is not only not fully portable between browser rendering engines, they aren't even fully portable across GPUs, as differing GPU restrictions like maximum texture size, texture memory, et al, impose different limitations on the maximize size of composited regions, how many can animate or update at once, etc.

You know, it would be cool if the text in your webpage took more than an awkward 30% width. And putting it in a bigger font size would be the shit, most definitely. Just saying :)

That said, this being butt-hurt about lack of adoption for your pet framework is just fucking overreacting. Yeah, I know, it's your shit, you made it, you're very proud, now it's personal. Bluntly: emulating a recognizable desktop interface in the browser is a shit idea. Your Mac OS emulation on a web browser over my Windows is the web equivalent of James Bond in pajamas. Not to mention that it completely destroys brand identification, and makes legal departments cry.

You wrote some cool stuff, it got you some fame and a job at Google. Stop whining.

Your day job is to build a piece of web software and you can’t take a few days to learn the ins and outs?

Do you know how many frameworks people are pushing out there? It's a mountain. And everyone thinks the secret to success is to be behind the latest greatest framework. Which the world doesn't need.

I know why they do it: because being behind a framework is an excellent point on the resume and will help you find employment, and so many employers really suck at evaluating people. But in the wake are a whole bunch of frameworks that some other employer is going to say "oh, you don't know CaffeineInjecterrr? Sorry, we need someone with 2 years of experience with it."

One reason founders complain that older people don't seem to know about the latest framework is that we've seen dozens of these things come and go like flashes in the pan.

I'm sorry if this comes off as a rant against this guy's framework.

  Your day job is to build a piece of web software and you can’t take a few days to learn the ins and outs?

This is the opinion I had when all of the previous Ember flare ups happened. My thoughts were exactly this. It would behoove so many of us to simply spend more time with new technology and ideas before we simply throw our hands up and say "it confuses me".

"Confusion" is not the word I would use.

Metaphorical "you" alert, not directed at parent:

You have a new-fangled thing for everyone to use. Scratch that, everyone who's been in your shoes, with your particular pains. Great, you solved a problem. Take some time to explain on your blog or in your GitHub README.md the context of your solution. "Because I can!" is a fine reason, but I don't spend time exploring something that was built because it was possible to build it.

So your new-fangled thing looks like it's a solution to a problem I'm having. I start to implement it and ... oh dear $DEITY, it's a mess. I don't mean it's ugly, I mean it's completely inextensible. It's a specific solution where a generic one would suffice. Or worse, you didn't think outside your little box to discover that this "fix" will debilitate 95% of the web servers on the planet if they were to try this solution.

My advice: if you build it, A) explain why, B) learn to build it well.

Fair. I'm only suggesting that before we collectively steam roll an open source project and its volunteers with "I don't get it, so it sucks" mobs, that maybe we spend a weekend with it. Maybe over that weekend we learn "oh dear $DEITY". And maybe, just maybe we contribute back to it in an effort to "help".

> This reaction to Ember just baffles me. Your day job is to build a piece of web software and you can’t take a few days to learn the ins and outs?

Couldn't agree more. While I do think that new technology should be intuitive and easy to use, you have to take the time to explore, learn and familiarize yourself with it. I think peoples' attention spans are way too short these days.

Time is money, and the larger your team is the more that the costs of learning a new tech are amplified.

Using his off-the-cuff, totally subjective "a few days" estimate, and assigning 'few' at a conservative 2 days, that would mean my team would spend roughly 50 days doing nothing but learning this one very specific technology. That's 50 days of not producing revenue or otherwise addressing client concerns.

Now, I'm not saying that we shouldn't be able to take time to do things like this (learn and embed new tech), but to dismissively say "a few days" like it's trivial is severely underestimating the impact of dropping everything during that time to do nothing but learn about a tech that will most likely be obsolete or superseded in only a few years time.

Web apps != desktop apps, at least not for now. When I looked Cappuccino a while back, I was impressed by what it could do inside a browser in terms of GUIs. However, when I researched a bit how I could use Cappuccino to solve the problems I was having at the time, I quickly got stuck.

A common use case for web apps is to allow for collaboration and/or sharing (e.g. Basecamp, Github....). When you look at the tutorials for Cappuccino you can quickly find out how to make an image editing app, but if you want to find out how to do even basic stuff (for web apps) like saving data to a backend, you have to search really hard.

In addition to that, Cappuccino always looked somewhat immature - even now the home page is somewhat broken. You don't want to bet on a new framework AND a completely new language when it looks immature.

This is a big part of why I left Web dev, too. I'm just sick of working with amateurs. When I'm at work, I have things I want to get done. This isn't the classroom, my teammates' lack of comprehension of even things as simple as dependency injection and MVC is their fault, not mine. My productivity and coding power shouldn't have to be hampered just because I'm working with a bunch of code-until-it-works-sorta hacks.

I'm so relieved to hear that there are no amateur native developers. Gives me a warm, fuzzy feeling inside.

The point seems to be that the web is cool and hip, which attracts people without computer science backgrounds. Without a solid footing in software design fundamentals, "hack it until it works" seems like a perfectly valid way to run a development cycle.

The desktop is not cool or hip, so the people doing it tend to be people who really want to be there, not chasing trends. It's a vast oversimplification of the situation, sure.

I think a fair number of us non-cs-ers end up in web development primarily because html/css is more accessible. I, myself, was more interested in design than coding, and I only learned the basics as a means to creating the designs I wanted. From there you might start learning some JavaScript, you might start learning some frameworks (like Backbone, or ember), and you might get fairly proficient with them... but I don't think the web attracts non cs people because it's "cool" or "hip".

I don't know about anyone else, but I learned almost nothing about software design when getting my computer science degree.

CS degrees are not a catch-all. Most teach data structures, algorithms, a little AI and database theory. Few teach how to structure programs.

I'm pretty sick of the "I work with people without CS degrees" excuse. Good software design really comes down to experience, persistence, and taste.

Necessary but not sufficient. I've worked with a lot of very experienced, very persistent people, and I'm sure they even had good taste, but they were shitty coders. In fact, their persistence was a requirement for them to be able to get their jobs done at all, given how bad they were.

Oh, the MVC pattern again. Here, take one of these: http://me.veekun.com/blog/2013/03/03/the-controller-pattern-...

You have to know how MVC works before you get to decide not to use it.

Nobody uses Cappuccino because it is written in such an obscure language making the entry barrier too high.

While frameworks like AngularJS won because they tried to embrace web technologies instead of replacing them.

"This reaction to Ember just baffles me. Your day job is to build a piece of web software and you can’t take a few days to learn the ins and outs? Presumably you’re in a large team because this project is important and will take some time to complete. "

What the hell is this!?

Your day job is to write software, perhaps, but you're going to come across Ember/X new web framework/whatever in your free time first. You're going to try it because you've seen the name floating around and it sounded cool. And - discover that it's impenetrable. That it acts like it wasn't designed for people to learn. That it starts frustrating and stays that way, when you were looking for a fun new project that moves fast and get results. Why would you keep going? Why would you give it time and then force your teammates to do the same thing, when the developer couldn't be bothered to make their software accessible? Why is it worth your time, as a human, to waste effort because someone who wants you to use their software couldn't be troubled to really try?

I don't understand what people are thinking when they create a brilliant, complex piece of software, and don't dedicate concerted effort to making it the easiest thing to learn ever. So - you want people to use your product, but only if the have to suffer? Or - you and your team have talked and decided you want some users, but, you know, definitely not too many, so you make it take a high activation energy to get started.

Software lives and dies by its barrier to entry and documentation. And far too many projects have far too little of both these days.

*I understand Ember is improving, especially after that post. Good! I haven't actually used it; I'm just going off what I read the other day.

To me this feels a little bit of a stab at the community for not adopting cappuccino but I remember what happened. Several years ago you showed us what it could do. We were impressed, we wanted to play with it but no, we weren't allowed. Instead of opening it up to the community it was kept locked away. By the time it was released (if ever?) we'd moved on. Now we don't care, we have frameworks that are good to work with (like angular). File size has nothing to do with it.

Author here:

First off, thank you all for the comments! Also, thank you Paul for posting this and helping me articulate my thoughts while writing it! Finally, thank you to everyone who has wished me success in my future endeavors.

Although I expected this post would step on some toes and anger several people (although I tried my best not to), I'm happily surprised by the reaction in general. HN tends towards the negative, by the Twitter feedback has sparked some great discussion.

I can't respond to each thread, but I'd like to take a moment to address some misunderstandings about the post. First off, I'm not trying to sell you on Cappuccino. The project as just as strong as it has ever been and the people doing the day to day work are very dedicated. I'm also not upset that Cappuccino wasn't a massive success or that it didn't dominate the world. The use of Cappuccino in my post was purely to illustrate my point of view. We built some great software, and I'm really proud of that.

The post is more about the stagnation of the community as a whole.

It's very difficult to write this kind of post without sounding defensive or like I'm bitter because my projects weren't massively successful. I tried my best, but ultimately you read the result. The tone of the post should read more as disappointment rather than anger. I'm disappointed in what hasn't been accomplished in the last 3 years. Many of you saw how great 280 Slides was when it was first introduced, and it's regrettable that it's not available today.

I want my friends to be very successful and I have no animosity towards them. The tools exist to build great things on the web and I hope this post inspires someone to make use of them.

Many of you disagree with my assertions, and that's fine. But I'm not the only person who feels this way. Paul Irish an avid web evangelist and Tom Dale of the Ember project share my concerns. And while we might disagree on the role of the web, I think many people agree that it should empower developers to build great things.

I've seen a lot of really great developers move on to other things because they share my frustrations and disappointments.

… and yes, the bits vs bytes line about dial up was my mistake. :) but I don't think it really changes the point.

[edits for clarity and spelling]

While I agree about the "no reinventing the wheel" part (after all that's one of the benefits of our product), I think Cappucino had made some choices that the random developer could not come to terms with.

I don't think that this is due to laziness or rouded-corner cravings (although the "new shiny toy" effect is very much existing).

I think this has more to do with the different timelines web developers face when building apps. Not all of them can afford the long learning curves required by Cappucino and the fact that objective-J is only useful for Cappuccino. Not all of them need to build app-looking apps or spend lots of time bending the defaults to look like something else

So the web community turns to low learning curve libraries as a side effect of not having the time to commit (as opposed to not wanting), better ones coming out all the time (things moving at a whole order of magnitude faster than in other software development areas), and using only 10% of the power brought by cappuccino.

This encourages more quick, single feature libs, which in turns brings more developers into that worldview.

Note that this argument is not about wether this is right or wrong. More about the why.

One can note that making those micro libraries work together is getting harder and harder, and that some folks (full disclaimer, we are one of those) are building glue layers and light architectures to guide developers and ease integration and make those libraries interact together seamlessly.

Take http://github.com/aurajs/aura (that's the one we're working on). It acknowledges both the fact that small, single-feature libraries are the way the web is heading, and that making those interoperable, and easy to load (a clean module system is central to aura) needs work.

Maybe small, badly written, non interoperable, single use libraries are just a step on the way to a truly modular, flexible and light web development process.

You and all the really great developers are welcome back to the web anytime you wish. Take a break. Enjoy your new job, and hopefully you'll gain new insights and ideas -- and share them...maybe even in new web technology. It's called burn out, and all great people experience it.

If you truly love the web, it has to be unconditional, just like real life. 280 slides is gone, just like Google Reader is gone, and somebody out there will fill the gap if there is one.

Web technology may be somewhat stagnant (I say it's "hindered"), but look at where the focus on technology has been since about...oh....2008. That also means there's opportunity ahead...you just have to see it. Calm before the storm, man. Now is the time to build and prepare. There won't be time to build solid technology once the next "sprint" comes.

Right before the next wave, you will stop hearing people say "web and mobile."

P.S. I learned Objective-C because I needed to. It was essential to my goals. Objective-J/Cappuccino never filled a need for me or my goals. It was as simple as that...with no emotions attached.

If there was no iOS, objective-c would be a niche language. People learn it because they have to, not because they want to.

The best of luck to you in what comes next.

Personally i don't really agree with this perspective of a stagnating community. There was never this age where the majority of web apps were properly engineered, in fact they used to be much worse, and before that desktop development was an even bigger mess (visual basic, anyone?). I see a definite progression. The current "average" architecture is some sort of client-side mvc layer on top of a set of web services. That's immensely better than yesteryear's php messes.

Thank you for writing this. You said about 90% of what I've been thinking about the web development community, and especially the client-side JS community for a while. So many of these problems have been solved, and a lot of them were solved with Cocoa, the iOS SDK and it's approach to building rich, complex user interfaces.

But, a lot of the js community is simply childish about it's approach to web development. They don't like static typing, they don't like the obj-c syntax, they're OBSESSED with micro -- they hate a framework (just use backbone!). Yet, as you point out in your post, so many of the problems we solve over and over again on the web were solved years ago by cocoa. But you know, they don't need all that "complexity". Please.

I'll stop now before I get a little too hand-wavy :)

The modern js developer is something akin to the antithesis of the architecture astronaut -- the architecture minimalist, or maybe the astro-hipster ;)

Thanks for writing this man, looking forward to see what you build next at Google. Thanks for your contributions to Cappuccino and other open source.

Conservative part here :). What's good about polyglot programming? Abstracting JavaScript with Objective-J? Or even abstracting HTML with JavaScript? I do not understand these tendencies to bring desktop software development practices to the web. It is different environment. You will never get to level of precision when hacking prebuilt widgets.

On the other hand I think that if the front-end is written in HAML instead of HTML, Coffee Script or Objective-J instead of JavaScript will not make application any better. It's often good usability and backend services which matter.

Stagnation of the community ? The reason why I did not consider about cappuccino is that I heard of another "new awesome framework which solves all problems" three days later. I would be delighted if the community could stagnate so I can catch up. You told about "reinventing the wheel", and that is exactly what you have done with Cappuccino no matter what you were thinking when designing it.

It may be great stuff, developers have no time for this. I can't afford losing 2 weeks of work learning a new framework which hypothetically makes me a bit more productive. Micro libraries are not a "trend" and file size is not the problem. We just want to know quickly what is behind the tools we are going to use and why using these tools matters.

I've been feeling similar to the way you feel for about 9 years, but with Dojo rather than Cappuccino. I stopped worrying or caring about fighting trends I cannot control years ago because it doesn't matter. I focus on building a toolkit that people use to build apps that require solid architecture and APIs. If I spent all of my time listening to or worrying about every critic or every person that doesn't get the power of having something more robust, then I would never make progress and would fail our users. You're not alone in your perspective. And yes, we were indeed all impressed with 280 Slides when it first launched.

I tried the Aristo showcase (http://www.cappuccino-project.org/aristo/showcase/), and it took 20 seconds to load on high speed internet connection w/ Chrome browser. I can understand developers' reluctance to use such a framework. If you've marketing a web app developed with Cappuccino, you're going to lose a lot of conversions with that kind of wait.

It's an app, I have to wait to download apps.

In fact I often have to wait for Gmail to download too.

They're just not framing the conversation well with the default loading screen. If it said something like 'this will only take a while once', it wouldn't be so bad!

I'm not sure this is a good objection.

Is this really written in Cappuccino? I get some crazy graphical glitches when I click any of the buttons or dropdowns on that page. I hope this isn't the quality they were talking about, but I will admit that looks like it belongs in a Desktop app.

40 seconds on my MacBook Pro.

Most of the time was spent downloading "AppKit.sj" which is ~2.6MiB, though, and my internet is pretty slow.

It loads in 5 seconds for me but the interface has many bugs and fails to implement many behaviors of native elements: scrolling with mouse wheel doesn't work, using the elements with keyboard is impossible because it seems many of them doesn't even have a focused state. When trying to focus on elements with the Tab key it's messing up the whole scrollview because after that elements aren't working even with the mouse. Pressing the Enter key redirects to another page. So it's a pretty bad showcase and the other demos too have the same issues. (using Firefox on Windows)

And, I've officially been called out in a blog post.

When I was considering SproutCore/Ember, I was the most senior person on a severely understaffed team at an ad company where turnaround time was measured in days, not months. A couple of days of research was definitely considered expensive, time-wise, in that environment.

I like the idea of continuously striving to better yourself, and I'll probably try Ember on a personal project to that end. However, adopting a new technology comes at a cost, and my team couldn't afford to spend what could easily have amounted to man-months of time relearning how to do things on the promise that we might be more productive in the future.

I really liked Cappuccino. Great features and 280 Slides was/is awesome. I follow its development for quite a long time.

Why didn't I use it? - As a complete noob, I found it hard to start (despite having good objc knowledge) - I checked the objj github language page, and didn't saw a lot of activity - Last but not least, Atlas, a Cappuccino IDE was supposed to be released.

I tried contacting Tolmasky about where the language & IDE were headed but never got a response.

"Never got s response" was pretty much the story of what happened to the (not free) Atlas beta. I suspect a certain number of people were just turned off by the way that was handled. (Yes, I realize that Cappuccino is a separate project.)

Yeah, I'm sorry about the way it played out too. Unfortunately, a lot of that was out of our hands.

Just curious, what happened?

I still think Atlas is key to developing in Cappuccino, unless you find another way to autocomplete ObjJ code efficiently.

Cocoa and ObjC only works because XCode does a lot of the heavy lifting for you.

Done with the web, moving to Google to work on Google Now; an app that relies heavily on web services to provide information to users...

I don't know why but I don't see that as being done with the web. I see it as changing jobs/projects, but still within the web community.

And also I don't agree with: Meanwhile, the web community will continue to solve the same non-problems over and over again.

The web community is larger than the people referred to within the article and I don't think it's OK to generalize like that. In the last 5 years the web has evolved massively and I don't agree with the fact that we are solving the same non-problems. Progress and innovation is out there, on the web, within the web... you just have to look from a different point of view (and occasionally sift through the BS).

I had a quick look and cappuccino apps are not crawlable, at least the demos aren't. Sorry, but it doesn't look like cappuccino gets what the web is about.

As somebody who uses text interfaces primarily I don't think desktop apps are amazing anyway, even if I see that they are better for some tasks.

The web browser is the only graphical tool I use daily, I don't need things inside to look like desktop apps. shudder

Yeah, this was my reaction as well. I'm kind of surprise someone has fondness for Desktop Apps anymore, because I only have one and it's my web browser so I do development and get other work done. Other than that it's terminal based running non-gui applications. Desktop doesn't really have a place on the web from where I'm sitting.

I don't think either of you are representative of the vast majority of computer users.

> Google Now is one of the most exciting pieces of technology I’ve seen in a long time. It’s solving problems that couldn’t be solved four years ago.

Showing people sports scores and traffic reports before they ask for them? Well, at least that gives us an accurate picture of your perspective (or lack thereof). The Galaxy Nexus is the best phone I've ever used or owned, but I couldn't possibly care less about Now. It's usually known as "that annoying thing that keeps popping up in my notification bar".

Everyone in the IT industry understands the frustration of trying to help people who are unwilling to change and insist on repeating history. Most of us don't give up quite as easily though...

That's shortsighted, this is what Google Now, can do.. well, right now. But in the future it will be extended to make your life easier, i.e tell you when you get close to a cinema that is displaying a movie of a director that you like, telling you that this near theater is performing a show that you really liked 2 years ago, etc.

These are not exactly stellar examples. The ones that I've seen that are good are:

  * Let you know that meeting you have will be tough to make it to due to traffic/you haven't left yet.
  * ...

Ok, so there is really only one good use case I know of. The others are items that, if I am not searching for it, I don't think I really want to know about it.

So, can you give some good expansions of that etc?

Your (lack of) use of Google Now is unrelated to the difficulty of the problem.

Predictive analytics is an interesting engineering challenge, and is probably what appeals to the author. On the other hand, re-solving the same web problems repeatedly is not.

From a product perspective, I can't see how it is exciting or new. Maybe there are some fancy gears under the hood, though.

I don't have it on my Nexus 4. If I left on all the notifications apps on my phone, I would be getting one every 2 or 3 minutes. I don't need another.

I find some of Google Now's features pretty awesome, perhaps creepily so.

I get an email with someone's itinerary, and on the days of his flights, Now shows me all the up-to-date flight schedules.

I look up how long it'd take to get somewhere on Google Maps on my PC, and in my car, Now is ready to navigate and tells me how long it'd take.

Lots of little things like that do add up to something useful.

I also think it may not have been possible 4 years ago, simply because it relies on a deeper level of spying on the users which didn't exist 4 years ago.

Amen brother. A lot of the problems stem from the cargo culting of various technologies and patterns (I'm looking at you, inflexible templating languages).

Also a lot of these issues come from looking at each piece in the development cycle separately rather than holistically. Frameworks like meteor are starting to address this, but we still have a lot of work to do.

I'm done with people making a big deal about new technology when old technology does the job better. People like this guy are spending their lives trying to figure out how to replace perfectly good native applications with ones that live inside web browsers. It's depressing.

It never even reached version 1.0.0 https://github.com/cappuccino/cappuccino/commits/v0.9.6

I'm not a web developer, so maybe I just don't understand, but why is it so important to the author that there be wide adoption of Cappuccino? Why does it matter to him that the majority of the web dev community is favoring "micro JS"? Why does he care that other web devs don't want to take a day to learn ObjJ? Why would he leave a path he's so passionate about just because other devs aren't being good software engineers?

Objective-J and this insane Cocoa worship? Are you serious?

Try to understand that the desktop and web are two completely different things. I would not want a UI that's designed ground-up for the desktop on my web browser. It's just wrong. I haven't seen Cappuccino myself (I mean, who has?), but stuff like GWT is just broken. HTML/ CSS/ JS exist for a reason: they're simple, elegant, and products of many years of evolution. Sure, there are cross-browser problems, but we're getting tools to deal with them.

Evolution doesn't happen by throwing away existing implementations out of the window, and starting with a fresh slate. Tiny incremental changes that preserve (atleast partially) backward compatibility always win. That is how software engineering (and engineering, in general) works.

Sure, there are a lot of incompetent programmers with the span of a housefly in the world, and a lot of "web people" are people- what does this "general truth" have to do with anything? More importantly, why does it annoy you?

  > The community has tossed out everything we’ve learned from the last 30 years 
  > of building applications in favor of being able to put something together  
  > quickly with complete disregard for maintainability or extendability.

I wonder, heretically, if maintainability/extendability is less important today... in the past, you'd build something, and then need to maintain it for years or decades. But today, if the pace of change really has accelerated that much, perhaps things change so much that it is easier to rebuild from scratch each time (instead of maintaining/extending) - even if you did carefully design it.

Is is just a small but prominent group doing web development (especially with ruby on and back-end and a front-end focus)? Is it just that we're still seeing industries adopt to the web, and old approaches slowly dying (i.e. in the midst of revolutionary times)? Or is this a permenant acceleration, that will only increase?

The web is the only future of software development. It's open, free, innovative, and unpredictable. It didn't pick your Cappuccino, sorry for you, but that's OK for the rest of us.

Something will come along and replace it when we finally pile too much on something that wasn't designed to do UI in the first place. The questions are "what" and "when" not "if".

Computers weren't originally designed to play music, video, or games. Yet, here we are.

That's a false analogy. The original assertion was about the underlying technology being used, not about what they deliver.

... yes well vacuum tubes and weaved wires as memory were replaced weren't they.

> Ember [being 49kB in size is] still far too big for many. Yet we conveniently forget that loading just a few images will quickly surpass that 49K of code.

50kB of compressed and minified JS aren't the same as a 50kB JPEG. The JS must be parsed and compiled which is a far heavier operation than decoding an image.

The last time I looked, it took about 100msec to parse and compile 100kB of minified and gzipped JavaScript on a desktop. However, after all of this you still won't get the full speed. It needs to run for a bit and then the hot parts of the code will be replaced with more optimized code. Again, this takes some precious resources.

An image, on the other hand, is decoded within a few msecs and that's the end of that.

Size is just one metric. It won't tell you the whole story.

Extremely good points in the article.

The front end "developer" community is awash with web designers who know enough jQuery (not Javascript, jQuery) to sprinkle some woo-woo on "websites", but who have no software engineering skills to create genuine rich, immersive applications.

And they're resentful because rich, immersive applications that require some engineering power, and some necessarily complex APIs are the way things are going.

And these people are noisy and influential.

But I think that's changing. We are going to have properly engineered, web mediated rich UIs. Not a mess of cobbled together freeware.

I often wonder if the problem with programming (or software engineering, if you prefer) is that we're still in the stage of where builders were in the middle ages. We assume the model of a "master builder" that has to know every aspect of every part of the process of building something new. Compare that to today where trades people specialize in one part of construction and hone it to a fine point. How many of us would hire an electrician to fix our plumbing, or a roofer to hang drywall? Maybe in the future this will be what happens to software development: you'll bring in a UI trade to implement the UI, a database trade, a caching trade, etc. People who've taken years to fully understand their niche of software construction very well.

The other issue we don't have to face is any governmental enforcement of building codes (yet). When you install wiring in a building, you can't just run it however might be "cool" or "new". There are a variety of rules, that come from best practices developed over many years, that are followed. Imagine if before a client or a customer would accept a finished app if it had to pass inspection by a 3rd party for compliance with these "codes". I think we'd see a lot of the current state of development fall away pretty quickly. The newbies who won't take time to learn their craft, or quickly jump from new toy to new toy without learning any very well would be under pressure to change or leave the field.

It's a shame to see someone who has been so passionate for a product reach this level of frustration but it feels like there's a lot of blaming 'everything that's wrong with the industry' for lack of uptake. If no-one is taking up on your offering, no matter how cool it is, then you need to look at what you can do to entice people in that you aren't already.

I've been back to look at cappuccino since the 280-slides first went up, and the get people hooked documentation has always been severely lacking (which is the case with a lot of projects, but it makes some of the complaints about peoples efforts to learn stick out Their docs have long had, to my mind, a tone of "it's just like objective-c, which is easy, so learn some of that and you'll be fine" [1] It does have pretty complete API docs, but they're not a lot of use without much deeper examples given than the few they have on the site.

When it comes to having thrown out years of software engineering know-how I think there's possibly something to that, and it is currently being picked up on in more recent frameworks but perhaps, as is being highlighted by things like backbone and angular, trying to shoe-horn dogmatic desktop software patterns like MVC into building apps for a relatively young platform like the browser is only a organisational stop-gap on the way to better, more suitable patterns that are still to come.

[1] http://www.cappuccino-project.org/learn/objective-j.html

So because I didn't embrace your framework due to its use of an unproven, unfamiliar language, you're going to hold your talents ransom from me? Maybe I made the right choice in not depending on Cappuccino in the first place...

Who ever said the web is the forefront of game-changing technology these days? It's just a platform that's fast to iterate on; that's why I like it, anyway.

Don't pigeonhole every web developer out there into some shitty imaginary stereotype you have just because most of us didn't use your framework.

"Almost every decent developer I know has no problem spending a weekend learning some new and cool tool, but sadly, this spirit seems to be absent from the web culture."

Hrm... the majority of web developers I know are constantly trying new tech, but they aren't just limited to front-end js. They're learning JS frameworks - trying out knockout, angular, yuk, dojo, jquery, ember, handlebars, mootools, extjs, gwt, qooxdoo, zk, and more. Plus they're learning about and using server-side caching, security, performance, server admin, graphic design, version control, search technologies, mobile development, database work, in addition to any language-specific stuff they're dealing with on a day to day basis.

I'm sort of sick of this attitude of "if you can't spend 2 days learning X, you're not a real developer". Or, more precisely, "web developers" don't do this stuff. I'll tell you what - if I spent 2 days on your latest framework, then asked questions about 'how do I do XYZ?', I'm more likely to get told off by self-appointed kingmakers that I'm not worthy of using your framework because I don't have the capacity to think at your level, or I'm not willing to commit enough time to learn it, or whatever.

So..., as they say in England, get stuffed.

He's going to work on Google Now, which is (more or less) the same thing I was working on with my previous employer. I could complain about how they are making a worse version of what we did, but I'm not going to do that. I'm just glad people are starting to see the usefulness of location-based development.

You may want to consider it from that perspective. People are getting interested in client-side frameworks. You laid important groundwork for that. Why not be happy about that?

There will always be a cycle of change in software development. Trust me, you will not find that this pattern changes switching from front-end JS work to backend "native" programming work. The only constant in this world is constant change. If I was not satisfied with that, I would be into something like architectural engineering. When you build a bridge over a river, it lasts for hundreds of years.

That said, I like a lot of the points in the article. I too have noted the absurdity of people complaining about how a 49kb file is "too large". I share the profound confusion of the hyper-simplicity movement, where we trade ~30kb of code for our ability to maintain software, which is a much bigger problem than performance in our industry today.

I think you have to make a few distinctions that recognize both different developer backgrounds and developer aspirations. It's like one of those "magic quadrants" that analysts love:

On the horizontal axis you'd have developer background: HTML/CSS/JS - OO Programmer On the vertical axis you'd have target application: Web Page - GUI-like App

Web pages are about text, color, images, layouts and maybe animations - in short "appearance". Page breaks are ok. GUI apps are about controls/forms, lists, tables, trees, to interact with a backend - in short, they're about "functionality". Page breaks in a GUI app are disruptive and are therefore avoided.

HTML guys hate latency, they're used to the most direct effects. OO guys usually don't mind to invest time before coming up with effects. HTML guys often shy away from learning a new programming language; it's important that they can apply the same skill set over and over again. OO guys don't mind to learn a new language, but don't like to miss out on paradigms and patterns they value.

Cappuccion would be in the upper right quadrant (OO Programmer / GUI App), while e.g. jQuery would be in the lower left ( HTML/CSS/JS background / Web Page). Probably, the crowd behind the HTML/CSS/JS profile is larger than the OO programmer crowd. And the amount of effort spent on web pages with spiced-up appearance is larger than that of carefully crafted rich internet clients.

But probably both have their relevance, so why trash one in favor of the other?! For someone advertising on the web a slick appearance is prime, while a vendor of set top-boxes values reliability of their web-based configuration GUI. The real problem arises, and that is probably Randy's pain, when the wrong technology and skill set is applied to a given problem.

This feels like a classic case of great product but poor marketing. The learning curve is a critical piece to adoption rates. When developers shun the framework, it's probably because it does not provide sufficient visible benefit for them to warrant going through the troubles of learning the framework. Once everyone gets onboard with jQuery, it's going to be hard to get people to switch over no matter how great Cappuccino simply because those people do not see the benefits outweighing the costs of switching.

It's also a chicken and egg problem. I'm going to go on a hunch and say there are way more developers who know jQuery than ones who know Cappuccino. Naturally, people would prefer using jQuery because it's easier to get someone else on board for projects.

I really loved the look and feel of Cappuccino and kudos to you for a tremendous effort! Unfortunately, I did not choose to use the framework precisely because I had to learn something called Objective J and for that, I'm sorry that I am part of the group of people who felt that the learning curve was abit steep.

I feel that the web culture in general is exciting, and often, moving almost too quickly for it's own good. To me, the service companies contribute greatly to the distaste people have for working with the web. Most are churn and burn outfits solely focused on the bottom line, which drive developers away from the web in search of a better fit both career and life wise.

Yet we conveniently forget that loading just a few images will quickly surpass that 49K of code.

It's not about loading time, it's about bloat.

This reaction to Ember just baffles me. Your day job is to build a piece of web software and you can’t take a few days to learn the ins and outs?

Sure but when there are alternatives which you can jump into right away with the same good results, why waste two days?

Exactly. The web is a very agile place. We don't all sit in cubicles and have weekly meetings. Some people just want to get things done, quickly.

Similarly, I've personally been seeing this a lot lately... the idea that "Done is Better than Perfect." And although I know it will never be perfect, the problem isn't perfection it's that this idea, fails to mention where the bar is set, or that it is set very low.

"Among his various possible beings each man always finds one which is his genuine and authentic being. The voice which calls him to that authentic being is what we call 'vocation.' But the majority of men devote themselves to silencing that voice of the vocation and refusing to hear it. They manage to make a noise within themselves...to distract their own attention in order not to hear it; they defraud themselves by substituting for their genuine selves a false course of life." -- Jose Ortega Y Gasset

The web is a very viscous place with all kinds of competition. Philosophically speaking, that form of competition can rob one of a chance at true mastery.

I remember a comment by a Smalltalker that pointed out many VMs are actually smaller than some images used on the web.

It seems like the author has been out of the workforce for several years, came back and realized this wasn't how things were when he left and now he's pissed.

The rate at which technology moves, you can't have any kind of framework that requires huge overhead, no matter how good it is. If people can't pick up and start building stuff with it right away, it's not going to gain traction. Add in decision makers don't allow you to just take a week to learn a framework you think will be awesome for a project you might be working on soon. Time is still money and nobody has the time or patience to learn something which has a high overhead. While I agree this sucks, it's just the way the industry works these days.

I do believe there will be a cost to this somewhere down the road, and the collective pendulum will swing back to how the author prefers it.

If the goal of "micro JS" truly is file size then yes, that's focusing on the wrong problem. To me the idea behind it is to have a bunch of little tools that do one thing really well. When those tools are combined you have something very powerful, flexible, testable, and reasonably decoupled. Not perfect. Nothing ever is. But better than having to use a massive framework and do everything the framework way in which case you're just working for the framework rather than it working for you. In my experience that is ALWAYS the case with a framework.

Put another way, Unix has been so popular for so long in part because it prefers the "a set of tools that do only one thing but do it really well" philosophy over a framework.

I share his frustration with the JS community. There is a rampant wave of a sort of "Ludditism" where people fight tooth and nail against innovations trying to make things better. The argument always comes down to "what we have is good enough, and OMG file size!!!".

I'm a bit more optimistic though. Randy and the Cappucino team blazed a new trail, and although he burnt out there are others who are continuing on and reinventing client side dev. Backbone was just a start. Ember, Angular, CoffeeScript, tons of other great tools/libraries are advancing things at a now exponential pace. Thankfully there are plenty of other early adopters willing to wade through the early buggy code to build projects on top of them.

Perhaps we've reached the point of too many layers of abstraction.

The web is designed for simple hyperlinked documents with some media elements. We've done a pretty good job of adding more stuff to it than that, but that doesn't mean it has proven wise or elegant.

My on topic, but only slightly, comment is that there is too much of an emphasis on framework these days. Why not libraries I can call at my discretion? It is me who is master of the code, and not the reverse.

tl;dr: Cappuccino is great, but the big crowd of frontend developers doesn't appreciate it, and insists on reinventing the wheels badly. That's why Randy is done with the web.

Randy, if you happen to read this: ignore the nay-sayers. There will never been a universal agreement on what the best technology is to get a job done. And there will always be people who try to troll your approach. That doesn't devalue your approach. Just build cool stuff with it, and be happy.

I'm with the OP on the 100's of badly implemented, slightly-sucky UI components. Cappucino did this brilliantly... but it was a too big paradigm shift.

Give me a beautiful UI library written with elegant and performant JS behind the scenes, but let me implement stuff in a non-desktop way using standard web technologies.

I like the concept of GWT, but Cappucino fell into the ExtJS/Sencha category for me - trying to reinvent the web using web technology.

All this careful superior engineering and all the sophisticated desktop UX libs and I can't name one desktop app that is nice to use (except SketchUp but that's thanks to innovative 3D tools not the UX which is ugly).

While many webapps written with crap tech thats only advantage is fairly flat learning curve look marvelous, responsive and are perfect in what they do.

Something else I wonder about is what this increasing focus on JS is doing to accessibility. Many sites won't even render with JS turned off; the no-JS version of them are only optimized for search robots - not human beings who rely on accessibility tools!

Is it just me, or is this so-called development moving us farther away from good accessibility instead of closer?

http://en.wikipedia.org/wiki/WAI-ARIA I'd lie if I said that I'm in a position to evaluate it's merits but it's there.

The mistake made here is the assumption that we know what the web is evolving towards: (ostensibly) something like the desktop. Hence the eagerness to do it all again but on the web this time. (Isn't that the essence of Cappuccino?) It would be absurd even if mobile hadn't happened. So why are we clinging to the same default assumptions?

Think again.

I see a lot of negative comments, many of them are from people far smarter than me and may be right, don't let them get to you. Their opinions are exactly that, opinions. You worked hard on something and it didn't work out the way you wanted it to, you have a right to be disgruntled. Congrats on the new Google gig!

I fail to see how the initiative to create decoupled components is a failure to grok "real" programming. If anything has come close to the code sharing vision of smalltalk (mentioned because objective-j is the strange cousin of objective-c, which is the step brother of smalltalk) it is the component approach.

you are done with the web community because its developers cant take the time to learn something difficult:

the web has produced as gateway to see programming work with little code. This will produce creators that do not have very much in depth experience/desire for hard problems. It also allows projects to ramp up and see results really fast. This is the trade off of the web ... deal with it or get out. You can be a teacher to these noobs or a nah sayer ... seems you picked the latter.

It sounds like you are surrounding yourself with poor programmers.

Google Now is not a framework its a product. Im sure youll be working with some web guys soon that will be implementing the web version of the product. Dont be surprised if they dont use the best tool ... things need to get done ... the most precious resource is time ... other things can be wasted before we let that one go.

Are you seriously telling web developers they don't deal with learning a new tool almost every day?!

How many damn Javascript libraries do we constantly integrate and learn? New css like languages (scss,less). Server side languages?

I'm sick and tired of people saying web developers are somehow 2nd class programmers who only understand HTML.

I remember to look at Cappuccino and say - WOW, this will change everything (again). But then suddenly nothing happened. No articles, no tutorials on peers, no visible discussions, etc. Things became worst when Motorola bought 280north. That was a deal breaker for me. Uncertainty about the library future.

This is not a good first impression for someone who had never heard of Cappuccino before. It reads like a verbose rage quit. It would have been more respectful to his colleagues and their (small?) community of users to explain his career decision while also remaining optimistic about the project.

Kind of surprised to see that this was posted by Paul Irish, considering he's such a huge web evangelist.

There are definitely some valid points in the article, but at the end of the day, it's hard for me to accept that The Right Way™ to do web development is to completely abstract away HTML, CSS, and JavaScript.

Certainly I live and breathe the web and do everything I can to advocate for its success.

I helped out Randy with this post because I think this perspective is both legit and underheard. Plenty of folks that work tirelessly to improve the web platform share frustration that many developers for the platform don't take it seriously as an application development platform, don't learn from past mistakes, and don't adopt the lessons learned from other platforms.

So honestly I think this painful truth needs to be heard if we expect the web to meet its potential.

Thanks for the reply. I was hoping you'd weigh in.

Why would I prefer to plug and play (and toss away) micro, modular libraries/frameworks designed to interop or to be independently used instead of using a monolithic technology that pushes to my throat ObjJ, Cocoa and a other contraints that follow (like IE8+ only)? That needs an explanation?

Awww, bless his heart. He forgot to add the #notsohumblebrag tag to his post.

Earlier today I was talking to another dev, and we were both wishing for a system like Delphi 7 for the web, with vcl and a vibrant ecosystem of 3rd party components (we were discussing how much SaaS leads to reinventing the wheel).

After 12 years in web dev, I feel his pain.

I feel as though everyone is leaving out a huge part of why the web has won. For the most part it's free. The world developed muscle memory that 'installed' cost money and web access was free. Not the only reason it won but a big one.

I think that this guy does not really understand how the web works. I mean, emulating desktop on the web has been tried many times and failed each time. He's right about some things (like people considering 49KB to be a lot) though.

I wrote this a little over a year ago:

http://lee-phillips.org/cappuccino/

Maybe not very kind, but just maybe people didn't rush to use this thing because they didn't inspire much confidence.

The end objective of web development is to bring desktop like experience? I wouldn't blame the failure of Cappuccino on the emergence of micro libraries. Cappuccino simply does not sound like what the web needs.

I am not a developer.Can anyone tell me why so many frameworks are being created?There seems to be five or six frameworks in each language,new languages being adopted to create frameworks.Why all this work?

Lets think about this from the opposite direction: Which projects on the web are the MOST successful? Facebook? Twitter? Gmail? Github?

Do ANY of those giants use big, all-encompassing super libraries? NOPE

At least Gmail does, perhaps Facebook as well by now.

Any time you argue that everyone is crazy except for you, one of two things is possible:

1 - you are in the wrong and have some biases that led you there. 2 - everyone is crazy except for you

Usually it is 1)

god DAMN, this is just about the most pretentious thing I've ever read. My face is making involuntary expressions of disbelief as I read more and more.

I just want to point out that Randy is advocating for the use of Cappuccino. Many people seem to be thinking that is the point. It's not.

So the author failed and isn't good at empathizing with people...

math0ne: your account has been dead for very nearly 5 years, this must be some kind of record.

The record might be that he was hellbanned for his very first comment.

There's no way yet another webapp framework, no matter how close it emulates the desktop, is going to gain lots of traction when it mandates users to use... Objective-J and JavaScript.

Seriously: being on your high horses and all is fine and well but your article is an oversimplification as to how the web works and as to what devs need and want.

For example lately I've been very interested in flapjax (FRP UI) and webfui (Clojure / ClojureScript with automated synchronization between the client and the server).

Functional programming and FRP UI may be the way forward. I'm not saying it is but I'm saying it may. What does Cappunico have to offer from the standpoint of determinism and reproducibility? Seen from a sufficiently distant paradigm, any web framework is the old, broken, inneficient way of doing things. Even your beloved Cappunico which you wasted so many time on.

Cappunico is a failure. It doesn't have to do with you being too intelligent and having figured out things earlier than others. It's just that there are many who devs do not want Objective-C / JavaScript and who want instead something really disruptive. As long as the one really disruptive thing doesn't come along, there shall be competition between lesser solutions (which Cappucino being one) Deal with it.

This, so much.

Randy's probably right about how silly all the tiny reinventions of UI frameworks are. And, though I'm generally unsympathetic to claims that a project is a failure because people don't want to learn the tools, I think it's safe to say here that asking people to use a language that's only marginally less horrible than JavaScript is not going to be a win from any perspective.

But he fails to realize that doing UI in a Cocoa-like framework is totally fucked (to be clear, I do Cocoa touch for a living; I'm speaking from experience).

I'd be making an understatement if I said I was unexcited about Cappuccino because it fails to even approach solving any of the truly important problems of UI work, which are (as you hinted) less about "can I make the table scroll quickly" and more about "can I decouple this monstrous mess of data, dependent requests, and presentation". As you note, FRP looks like a step in the right direction toward solving that problem. Traditional Cocoa-like approaches are a step backwards in that area, not forwards.

I read the post differently.

I think he was simply pointing out the fact that we've stagnated at the same level of abstraction in the web space for a long time. Cappuccino is just the example he's using to point out where we were relative to where we are and in that light it's almost a regression.

For the purposes of his larger argument, the way that Cappuccino delivered its abstractions (Objective-J) isn't as important as what it allowed you to do with the web.

Second this; while I'm not a web dev, I do bump into it occasionally, and it pains me any time I see someone re-re-re-inventing the wheel, often badly. I don't know about Obj-J or Cappuccino, but just the OP's comments about modules had me nodding in agreement.

Another related example, I recently learned of H2 (http://www.h2database.com/html/main.html) and I honestly had to ask myself, why is this needed? If you want an embedded SQL DB why not just use SQLite? If you want a full featured enterprise SQL DB, why not just use PostgreSQL? And don't even get me started about Walyand/Mir as opposed to Xorg, or the whole debacle of disabling separate /usr because of rewriting things like init to use glib and other non-base system dependencies. The hubris, lack of knowledge of history, lack of foresight, and just plain lack of professionalism (to put it politely) are staggering.

> Another related example, I recently learned of H2 (http://www.h2database.com/html/main.html) and I honestly had to ask myself, why is this needed? If you want an embedded SQL DB why not just use SQLite?

SQLite is native C, H2 is Java. There are significant deployment and potential stability advantages to not using native binaries in your Java software.

There have been some neat hacks to get sqlite running under Java, including using a tool that translates MIPS binaries to run under the JVM, allowing the use of libsqlite as 'native Java' code: http://nestedvm.ibex.org/

This is, however, not something you'd likely want to use unless you absolutely need to interoperate with sqlite's data format.

I noticed that the fully native Java was a big selling point on H2. Admittedly, I'm not doing much Java these days (slight understatement), but I was under the impression that the Java interfaces to SQLite were fairly mature? Top two links in Google go to a very informative SO answer with lots of options and SQLJet. Bad sign though that "apt-cache search sqlite | grep -i java" gives no results on Debian stable. I'll grant that fully native Java is a compelling reason in some use cases ;) I'm now curious what they use for SQLite on Android, as I know fairly well that it's used all over the place there.

H2, more than anything else, is great for testing. It can emulate oracle and DB2 syntax, among others, to allow us to reasonably approximate a running system on our own machines, then test against the big database (which is much slower) at a later stage.

Fully native java and an in-memory implementation mean that we can very quickly implement it in our systems.

I second this. We use H2 in our JUnit test cases a lot, because it is the fastest open-source in memory Java DB (tests are finished very fast).

This is my feeling, too. I do iOS development for living and whenever I have to do some web programming, I feel like writing in assembly, having to micro-manage things I should not care about.

It pains me that Cappuccino has received so little interest, for despite all its possible drawbacks it represents a very interesting path for future web programming.

> whenever I have to do some web programming, I feel like writing in assembly, having to micro-manage things I should not care about.

That applies in the opposite direction too. Native programming seems to involve jumping through insane hoops just to do simple things:

Like "put some text on the screen, this part is bold".

Or "open this web page".

Or "put these things next to each other, with 10px in between, unless the screen is a little narrower, then put the second one below with no margins".

Or "I want to play with the layout of this every day for the next month, have the new layout download every time someone uses this app".

I can write complete, ready-to-distribute web apps that do those things in 1-3 lines of code. Doing so in iOS takes many layers of indirection.

Look at something like Ember.js. The abstractions being put in place are fascinating. Or CSS. It's is an incredibly powerful abstraction on something that iOS just scratches the surface of.

The truth is, both the web and iOS have evolved to deal with specific pain points. Some things that are easy on one are hard on the other.

This idea that web programming is under-abstracted seems really narrow-minded to me.

Have we stagnated? Or has the rise of lower powered web-enabled devices forced us to re-think the classic rule of programming (ie that applications can progressively increase their system requirements as PCs are progressively becoming more powerful to handle the load).

Let's remember that even as recently as 2-3 years ago, top end phones and tablets were single core ARM processors clocked at ~1GHz. And devices of that spec are still in wide spread use even now.

The point of the HTML is that it's an open standard which should work on any internet device regardless of platform, patents and what not. While I admire the ambition of the guys who push the envelope of what the web can do, they often forget that everyday sites also need to cater for low end devices used by everyday folk.

This trend is likely to continue as the billion or so potential customers in emerging markets start getting their hands on affordable smartphones.

The idea right now for telecoms and device manufacturers is more about getting their stuff in as many people's hands as possible, not about whether they can replace a screaming desktop with a handheld.

Totally. And this is the right attitude as well.

Let make information available to everyone; then we can worry about making those sites fancier than a strippers underwear draw.

We have stagnated because new APIs, frameworks, environments appear at a rate too high for anyone to build meaningful things on top of them before they become obsolete again. We also wait longer until we adopt these new technologies, therefore narrowing the time frame to actually use them even more.

Low end devices are a good point, my main gripe with all these frameworks is that they make it very easy to break many conventions of the web (URLs no longer work as intended, extra efforts is needed to make the pages crawlable etc.).

We're stuck at the same level of abstraction because the platform forces us to. The base html/css controls are incredibly simplistic, to the point of being pretty much useless for building proper UI. Everyone is struggling with how to abstract their way around this fact.

You have solutions like gwt, cappucino and extjs which build custom ui components that behave properly, ad the infrastructure to tie them together. This works, but only if you never step outside the framework, and only if you're willing to deal with bloat, which sucks. Then you have frameworks which just give you the infrastructure, not the custom ui components, like ember or backbone. That also works, but because html's core controls are horrible the only way to get a great ui out of that is to build really custom ui, essentially rebuilding what the complicated frameworks already provide out of the box, which sucks. And then you have one-shot solutions that try to bring just one or a few ui elements to the web, trying their best to fit into the raw platform. That also works, but because the raw platform is so painful to use, it doesn't scale.

Really, the problem is that html/css/js is the wrong base for building abstractions on. I have great hopes for shadow dom and web components, but that's only the beginning of what we need. The platform itself needs to evolve a great deal to make building high-quality ui something that doesn't lock you into a vendor's toolset.

Personally i chose extjs and am waiting it out until the platform matures underneath it. I'd like to see the extjs codebase evaporate as parts of it get replaced by native code, until it's nothing but syntactical sugar on top of a rich base platform. That's going to be a while though.

Thanks for a fascinating comment! Do you have any opinion on the parenscript project? http://common-lisp.net/project/parenscript/

Seriously.

Is he saying that a product failing is the fault of customers? That's simply ridiculous.

No matter how you put it, one way or another it is ALWAYS you fault, I don't even know what it might be the problem, but it's your fault.

He can blame the technicians for being lazy or stupid, and he does it, but the fact remains, they didn't wanted to use his tech.

Some front-end devs are technicians that need to generate value quickly, and most can't afford investing time in a new complex system that may disappear tomorrow. If there are no clear advantages on the business side that justifies the investment, then there is no pressure to the front-end devs to do it.

Good engineering also means optimizing technical designs for the usage of the technology in order to reduce costs. That includes factoring in the skills of the technicians that are going to use the tech (and the cost of training them).

There's no way yet another webapp framework, no matter how close it emulates the desktop, is going to gain lots of traction

My objection is right here. No need to mix in the Objective-J at all.

Microsoft's ASP.NET tried to do this thing, to a surprising amount of success. The context then ofcourse was "Web 1.0" pre-Digg internet and the only other "big thing" being PHP.

Microsoft managed to take the whole "Draw a GUI in Visual Studio and it's a done app" thing and take it to the web, much like it originally defined Windows-programming with Visual Basic. It was an immense success and very cool.

Then Web 2.0 (as much as I hate that term) came and changed the game.

Then not working on the raw metal became an obstacle to making ASP.NET do all those cool things which everyone else was doing. It's killer feature (not having to know HTML) became it's Achilles Heel. It's abstractions became a hinderance to all the things you actually knew how to do once the framework got out of your way.

I may be a tad pessimistic and cynical, but I doubt a framework whose main goal is to give you a non-webby workspace will be a long term success when you're actually working on the web.

Microsoft learned that (too late?) and have tried to remedy the situation with ASP.NET MVC, but by the time that became useful, it seems lots of .NET developers went off to find cool stuff elsewhere.

A new framework like that, right now, seems like an utterly dead end.

Every time I see someone talking about using JavaScript as working "on the raw metal" I die a little bit inside.

I'm joining OP's web haters club, if he starts one.

"I die a little bit inside."

Every time I hear a programmer condescend towards other programmers for using a higher level language, I also die a little bit inside.

How is javascript not the "metal" of the web? It's a core part of the modern web experience upon which many libraries and frameworks have been built.

I don't understand why that analogy is bad and is worthy of your condescension.

Are you just being condescending to the entire web paradigm in general?

Well, to start, it's a scripting language, not a compiled language. At least on the web.

Javascript is to the web what assembler is for hardware.

Nope, it's (JIT) compiled. No modern browser uses an interpreter anymore.

JavaScript is compiled by every modern browser, in fact it has some of the most advanced compilers that are out there. And the dichotomy of "scripting" vs compiled languages is false. You may be thinking of dynamic and static typing.

PFFFT!

A language interpreted at runtime, by an application that runs inside an operating system (which may or may not be virtualized) that boots into protected mode. Let me get this straight, you're ostensibly describing this as being "bare metal"?

Have you ever heard the terms "kernel mode" or "ring zero"?

Oh wait, you qualified it as "the 'metal' of the web"... so it's an "analogy". Sorry to be pedantic, but when people say "bare metal" in the context of platform virtualiztion, "bare metal" literally means just that. The executable code is compiled into binary that directly correlates to the copper wires (and etched semi-conductors, or perhaps even vacuum tubes) of the specific, real, tangible machine intended to execute the code. JavaScript is nothing like this.

Maybe the packets and datagrams of transmission protocols are the "bare metal of the web", but JavaScript most certainly is not.

Check out the Communication Systems OSI Model.

http://en.wikipedia.org/wiki/OSI_model

I'd contend that nothing above Layer 4 could be considered analogous to "bare metal".

You're really reading too much into his analogy; it was a quick informal way to get a point across and everyone (should have?) understood what he meant.

> The executable code is compiled into binary that directly correlates to the copper wires (and etched semi-conductors, or perhaps even vacuum tubes) of the specific, real, tangible machine intended to execute the code.

You do realize it's been a while time since the ostensible instruction set and corresponding assembly language of most modern CPUs translated directly to the bare metal execution model of the machine. http://en.wikipedia.org/wiki/X86#Current_implementations Even on RISCier architectures, virtual memory, deep instruction pipelining, super-scaler dispatch, branch prediction, multi-layer caching, inter-core cache coherency, etc. introduce a huge amount of abstraction between the instruction you write and the actual execution on "bare metal".

On current chips, x86 machine code is as much a virtual machine as java byte-code. It's abstractions all the way down, you just pick different levels for different classes of work and "bare metal" is now just a label for one below where you landed.

>On current chips, x86 machine code is as much a virtual machine as java byte-code.

I understand what you're trying to say, but this is way off base. x86 instructions may not align exactly a processor's primitive operations, but that doesn't mean abstraction makes x86 similar to the JVM, at least not anymore than it is similar to a Bible printed on papyrus. But nobody cares if ancient Hebrew is just as much a virtual machine as Java, so stfu.

You should leave out the stfu next time. You comment stand on it's own without it.

Silicon isn't technically a metal anyways ;) A metalloid at best. http://en.wikipedia.org/wiki/Metalloid

"Sometimes writing markdown feels like working on the bare metalloid"

Ha! Okay, ya got meh!

But! IN THEORY You could still construct a processor out of electromagnetic relays (springs, contacts, leads and solenoids only), if perhaps your goal was to never exceed 100 Hz.

(...at first I was going to point out http://www.bigmessowires.com but it turns out that beyond the wire wrap connections, he's still using modern semi-conductor packages)

I think you're on your way to praising the god particle: http://abstrusegoose.com/98

The metaphor is apt. JS is a terrible terrible language (with some nice features). It's very easy to do the wrong thing. It's very hard to organize large projects and it needs higher level frameworks to be usable and portable. And with projects like asm.js it will be a necessity (not an option) to write in a higher-level language in order to generate performant JS web apps. So the "raw metal" metaphor is apt.

As far as I know, Microsoft's ASP.NET was kind of a failure. It produced Frontpage-grade (e.g. dirty) html. It did the wrong thing by default and didn't help much with the right thing.

It also brought in some clueless people who had no idea how HTTP, HTML and JS work, with questions like "why my button.hover event does not work" (this was pre-ajax, the event code was on the server and had no ways of knowing the user is hovering over the button).

Maybe it was OK for CRUD and internal applications but not for ordinary websites or sophisticated web apps.

.Net MVC is seen as a success: clean HTML, predictable things under the hood, jquery.

> Microsoft's ASP.NET was kind of a failure.

Eventually, yes. Which I addressed. But prior to the whole Web 2.0-thing, it was actually pretty successful by lots of means.

I'm not saying ASP.NET MVC is superior in most respects. There's no doubt about that, but it was clearly a late-started initiative which should have started much earlier. ASP.NET already had lost its respect and was a waning star by the time they got the first version out the door.

These days it's actually fairly pleasant to work with. The only thing I can say I really dislike about it is that you can still find the traces of all the old ASP.NET warts there, if you go beneath the hood.

Edit: Thanks for being the only one replying to the actual topic at hand, as opposed to doing inane discussions about what "real metal" is these days. And that's coming from someone with a degree in electrical engineering and design of digital electrical circuitry.

I think ASP.NETs popularity wasn't because developers thought it was cool, it was because Microsoft was very effective at selling it to management. What better sales pitch than to suggest you could whip up a "rich" business application with a series of drag and drops and a few tweaks to the code, in no time at all!

Oracle are doing similar things these days too.

You've nailed it. From another aspect frameworks and abstracting libraries are designed for reusable assets, paradigms, and resources. This is great if you want to create cookie-cutter websites, like corporate CMSs.

This style of thinking damages applications. Yeah you've made it easy to build a form with validation, but users don't need half the form fields, let alone the form itself. Also if you thought about it, you could have came up with a more innovative way to capture user data that would lead to more registrations, and thus more users.

He didn't even mention the Dojo Toolkit. Dojo has all of the challenges Cappucino faces, except that its all Javascript, as opposed to Objective J. Dojo has the awesome table grid, the widgets, all of that. I don't use it anymore, but it was pretty powerful on the project I uesed it on.

But... Google New Thing!

although it seems to be "yet another web framework", be careful when you try to reject it only because of that.

Web frameworks happen because people try to leverage their existing knowledge, skills or codebase in a web domain.

i.e. GWT & Vaadin are web frameworks where you use java and there is literally no need to know JS/HTML/CSS to create some really nice web apps that perform well.

I don't want to say that you don't need JS/HTML/CSS knowlede to build web app. But those frameworks allow people to be productive with almost no such skills.

"you don't need JS/HTML/CSS knowlede" But what would you do when it breaks?

Or when a change of behavior asked that is not supported by the framework out of box?

If you actually bothered to read his post, he was actually using ember.js as his example, which is decidedly not written in Objective-J. It's plain javascript with handlebars templates.

see you later

I'm sorry about the harsh tone of my other comment (most upvoted comment so far) but the tone of the author is aggressive so...

Now on Cappucino's website I read this:

"Cappuccino is an open source application framework for developing applications that look and feel like the desktop software users are familiar with."

and I can't help but think this is all wrong. There are many (most ?) users who are now familiar with webapps, not desktop apps. And what they want is definitely not a webapp looking like an old, outdated, Windows or Java / Swing app (which is exactly how Cappucino looks like).

What many user wants is webapps that look like webapps and if they are "forced" to use a desktop app, then they want a desktop app looking like a webapp (I'm thinking about the client-side desktop apps embedding a browser + JavaScript engine).

I'm also pretty sure that devs who have done both advanced HTML + CSS to "customize" the UI of their app do much prefer that over, say, the total and utter madness that Swing is (I'm speaking of what I know).

Yes, HTML and CSS have their warts... But you'd have to a masochist to want to create a desktop app using Swing.

Users I'm sure they don't want apps that looks like desktop apps: they now have browsers, iPad, Android tablets, etc. They hardly remember what an old Windows desktop app was like.

Devs I don't know: I've done my fair share of Swing and the last thing I want is go back doing that.

I kinda like this "simple / straight to the point" webapp worlds that we have right now.

Out of curiosity I just looked at the Cappucino website... the controls look modern to me, not like "old, outdated, Windows.. Swing" apps.

Swing is a particularly bad example of UI toolkit. Look at things like Qt, Gtk or Cocoa instead.

>Ember is very close to the “micro” side of things, but at a “whopping” 49KB (less than a second download on a dial-up internet connection)it’s still far too big for many.

Erm, isn't dial up at 56kilobits per second? It will take much longer than a second to download 49KB.

Wonder where it all went wrong?

> Objective-J

oh yeah

Objective-J is not the problem itself , the problem is that Cappucino is very hard to debug in the browser itself. I went through the tutorial , had an error , and there was no way i could know where it came from.

I dont have a mac right now so i cant use the builder , but i dont think that Cappucino failed because it is bloated. I use Haxe which is even more bloated than Cappucino. Yet Haxe allows me to debug a JS app directly in the browser with source maps , Haxe has a very smart and powerfull compiler.

Maybe it is matter of timing. But i think it is just a matter of tools. Again Obj-J is not horrible to work with. And Capp apps are definetly neat.

The disclaimer at the bottom of the page says it all:

"(Obviously these opinions are my own and of course do not represent the opinions of my future employer)"

He has no real world experience doing anything that matters so just chats about what he's witnessed from within his bubble.

In the interest of not being a coward and deleting what I previously said...

I just read his 'about me' section and feel a tiny bit humbled..... not massively though.

Reply on news.ycombinator.com