The Value of Open Standards

Tuesday, February 16. 2010

...or why HTML 5 will fail as replacement for Flash (at least for the next years).

But first let me make clear that I am not a fan of closed standards and closed source. I love Open Source and I use Open Source software wherever possible. However sometimes it's better if there is one instance that has the control not just over a format, but also about the interpreter and renderer.

With the announcement of Apples iPad a big discussion started regarding the non-existence of Flash on it. Steve Jobs responded. He claims that Apple won't support Flash since they claim it's buggy and responsible for a lot of crashes on Mac OS, but what really makes me upset is that sentence:

No one will be using Flash, he says. The world is moving to HTML 5.

Also in blogs postings that talk about this topic several people mention that HTML 5 will be the future and that no one needs Flash.

Bullshit! Will people never learn from the past?

The History of Problems With HTML

Some years ago CSS 2 was praised for being a solution to layout and styling nightmare we faced with HTML 3. But ever created a complex page with HTML 4 and CSS 2? And then tested it in different Browsers? If you did that and claim that you never had experienced difficulties due to different behaviors of different browser you're a liar. Because either your page was a really simple page or you had difficulties.

These problems are nothing new. They are existing since HTML 3 and became worse the more complex HTML got and especially since Microsoft did not care about the W3C standards and started to interpret HTML their own way. But also other browsers had their own behavior - intentionally or not. HTML designers are used to hefty workarounds to make a page look like it was intended in all browsers.

What about HTML 5?

A little excerpt from an article on Wired:

Opera and Safari have been pioneering HTML 5 support for some time, but Firefox and Google Chrome aren’t quite as far along with their HTML 5 support (although the coming Firefox 3.5 will close the gap considerably). Internet Explorer 8 is somewhat further behind, though it too incorporates a few HTML 5 features.

[Highlighting added by myself]

Does this sound familiar to you? Different browsers supporting different sets at the same point in time of a new HTML version?

How long did it take until all browsers passed the ACID 2 Test? 3 years! And ACID 2 is just about CSS. Take a look at the ACID 3 results: At the time of this writing there are just 4 browsers that pass the test.

It will take another several years until all browser are on par with HTML 5. Learning from the history we will never be able to see that HTML 5 applications must not be tested in a couple of browsers for their compliance and we will never get rid of notes like "This page is optimized for blablabla Browser."

And even when there are finally browsers available that support HTML 5, how long will it take until all users have switched to the current versions? Just have a look how many big organizations are still enforcing Internet Explorer 6.

Robert Reinhardt raised a couple of points why HTML 5 will not be ready to replace Flash in the next years in his excellent post. He talks about Flash Ads (whether you like it or not - it's a part of todays web), animated suplementations for a web site, video (see also Ben Schwartz' article about H.264 licensing) and games.

The Enterprise Side of Flex

However I am particularly interested in creating complex RIAs which seems to be forgotten most of the time when the Flash platform is compared against HTML 5.

Complex RIAs need much more that just static HTML and CSS. They require scripting and DHTML in combination with AJAX extensively. Yes, there are frameworks that hide the browser specific, ugly parts from you and serve as a high level abstraction layer - until you are forced to leave that layer, either because you need your own components or want to do something differently. And then you are still forced to test all browsers and adjust your implementation once again. Even if you stay on the chosen frameworks path - you have to test every single feature of your UI since the framework is most probably not bug free. Even worse is that your UI might break in a future version of a browser and you cannot test the future today.

So far I only talked about the runtime environment of a RIA. But there is more:

Enterprise ready IDE support with enterprise ready debugging capabilities: Adobe offers Flex Builder (next version: Flash Builder) and there are also excellent 3rd party IDEs like IntelliJ IDE or FDT.
The existence of profilers: Built-in into Flex Builder.
A typed language to catch programming errors early.
Frameworks for automatic testing - Unit as well as UI testing.
But much more important: The existence of a set of reusable, skinable and extensible control components like buttons, radio buttons, data grid and various visualizations like graphs or dynamic trees. Even better, if 3rd party vendors also jump in and provide much more components. Just launch the Tour de Flex to explore the wealth of just a subset of available components for Flex.

Today Flash is still seen as a plugin for providing flashy animations and multimedia capabilities in a browser. This is an area, were HTML 5 might compete someday in the future. But until HTML 5 can compete in the RIA space there is still a long long way to go.

Open Formats are not Enough

Face it! Fact is that today the Flashplayer is the only platform for complex RIAs that provides a browser and platform independent rendering engine with the necessary tool and API support and a platform that is installed out of the box on almost every single computer.

Yes, it's a closed platform and there are many disadvantages. Ask my colleagues - they know how often I ranted about Adobe. But on the other side there are also advantages. A closed platform can be controlled by a single organization, which ensures the compatibility across certain boundaries like operating systems and internet browsers. There are no fights with proprietary extensions between organizations with different objectives. There are no incompatibilities due to a different interpretation of a standard by different implementations of the document parser and rendering engine.

The sole definition of a standard description format like HTML to describe UIs solves only half of the problem. It's not the HTML 5 standard compliant document that renders the UI - it's the interpreter. Today you can have a perfect W3C validated web page. Your customers do not care as long as their own browser does not render that page properly in their Internet Explorer or whatever browser they use. Customer just want to have a page that looks like expected by the designer or a RIA that just works. A customer does not care about browser wars and that is his perfect right.

Other examples:

WSDL: Intention was that any framework can read a WSDL an automatically generate a client. Fact is that various frameworks produce their WSDLs that are incompatible with other frameworks.
OASIS Open Document: Works perfectly in OpenOffice, the reference. Several incompatibilities when you mix OpenOffice with KOffice.
SSL certificates: Minor glitches in the standard lead to this bug in Ubuntu that causes subversion to stop working with the libneon-gnutls library that performs a strict checking of the certificate in contrast to the older libneon library.

There are many, many other examples, where a open standard fails to provide the right customer experience. Open Standards are great, but let me repeat: The end user does not care. He wants a working application - period.

So when Apple says "Use HTML 5 to create web applications" they really mean "do not use Flash, because we don't want you to use a runtime that is not under our control - create HTML 5 apps optimized for Safari, but it's not our problem, if they do not run on other browsers than Safari." By saying that Apple surely wants to achieve applications that are running perfectly on their platform. After all they take the effort to review the applications before consumers are allowed to install these apps on their iPhones and iPads (interesting read: Apple's Mistake by Paul Graham). But they don't care at all about other platforms. So its the programmers problem to provide cross platform applications. Or even better for Apple: Make the programmer busy providing applications for iPads and iPhones and make it as hard as possible for him to port this to other platforms by using an abstraction layer like Java or Flash.

When Steve Jobs says "The world is moving to HTML" he has substituted 'Apple' with 'world'. Ist Steve Jobs becoming a megalomaniac?

Apple: "Flashplayer Responsible for Most Crashes on Mac OS"

By the way. The Flash plugin works pretty well under Microsoft Windows and Linux. I never had a crashing Flashplayer that even crashed the browser on Linux - even more the whole operating system. Linux is certainly not the primary area of investment for Adobe. So how comes that Linux is obviously better supported by the Flashplayer than Mac OS? Maybe because Apple does not really support (or even hinders) Adobe?

Wishing HTML 5 all the Best

After all I wish HTML 5 all the best and hope that I will be proven wrong in the future. I hope that once in the future I will shake my head and wonder how I was able to write such an article. I hope that I will feel like an idiot when HTML 5 has become the de-facto standard for RIAs with enterprise ready IDEs and browser independent rendering quality.

But I am very pessimistic that I will experience that.

Posted by Carsten Schlipf Filed under Thoughts.

Comments (10) Trackback (1)
Leave a comment or send a Trackback from your Blog.

Trackbacks

Trackback specific URI for this entry

Adobe: Creative Suite 6 not on Mac OS
Some thoughts of what could happen if that headline above could become a reality. Most probably you've already heard about Steve Jobs dislike of the Adobe Flash Player. A War that started when it was announced that the iPad will not support Flash and hi

Weblog: Occasional Thoughts
Tracked: Apr 30, 23:21

Comments

#1 Florian on 02/17/10 at 09:06 AM [Reply]

First of all, your post is a buzzword parade, I'm not, and never will be, interested in *anything* billed as "enterprisey ready", I have star trek next generation at home on DVD, thank you very much.

Testing, in general, should be an automated process, regardless weather you have to support one platform or many, and if it isn't, well, you know who to blame, yourself.

As to cross browser issues, I don't know what rock you've been living under, but those things got a *lot* better in the last couple of years, to the point where I can, (gasp), write a webpage, test it only on one browser, and it runs exactly how I want it on all major browsers (admittedly, if you've no clue whatsoever and no experience in html, and you don't know about the meaning of "don't use CSS/HTML hacks, ever!" then you might be out of luck).

Now in regards to your oh so shiny flash/flex wonderland. Let's face it pal, it sucks. It sucks because, well, I can't just sit down, bash out a couple of lines of text, be done, and have something that runs. It also sucks because it just fails at a couple of its core competencies (like playing back video, which is just mirred with so many issues I won't even begin to describe that train-wreck).

Now it's true, HTML won't supplant flash anytime soon. However, if you want to write RIA applications, you don't *have* to do that in a proprietary standard using convoluted tools. It's true, you won't get all of the vector niceties of flash, and you'll have to deal with the occasional cross browser issue, but these aren't killer issues anymore, they're not even major.

What I think gets you so rilled is that, despite all its messyness, and cross browser gotchas, and open standardyness, HTML is *actually* inching to a position where it takes on *some* of flashs core competencies, and consequentially, people use it for that. And I think that makes you scared, since you'd have to basically learn HTML now, of which you obviously do not have any clue.

#1.1 Alex on 02/17/10 at 09:18 AM [Reply]

"However, if you want to write RIA applications, you don't *have* to do that in a proprietary standard using convoluted tools"

Very true, but the amount of effort required to write a nice UI that works across the majority browsers is enormous despite the likes of GWT - better to use something like Flex or JavaFX/Java which is designed for this kind of thing, rather than push the square peg into the round whole.

#1.1.1 Margery on 07/08/11 at 08:24 PM [Reply]

Whoever wrote this, you know how to make a good artlcie.

#2 Alex on 02/17/10 at 09:08 AM [Reply]

Nice post: the RIA part is often overlooked when it comes to these comparisons.

Indeed, building moderate to extremel complex UI's in HTML/CSS/JS is very hard work and not really what those tech's are designed for: I don't think the new Canvas tag will do much to resolve this either - but I could be totally wrong...

Personally, I think that HTML 5 will trundle along, and, like you pointed out, will have pot holes and surprises for all developers due to different companies implementing the spec slightly differently.

#3 SM on 02/17/10 at 11:50 AM [Reply]

Each technology has its advantages in certain types of tasks!

#4 Danny on 02/17/10 at 06:54 PM [Reply]

"Flash works well in linux"? It routinely gobbles up 100% of the CPU and only in the last year or so has solved the issue of the video and audio not being in sync.

No, flash on Linux sucks, even if it sucks less than it used to.

#4.1 Carsten Schlipf on 02/17/10 at 07:42 PM [Reply]

Hi Danny,

well, sure there are configurations on Linux, where Flash does not work properly. But I can assure you that I do not have any problems at all on various Linux systems for more than the last 2 years.

At least it's not as worse as Steve Jobs describes it for the Mac, where he claims that Flash can crash the whole system.

#5 Darren on 02/18/10 at 12:58 AM [Reply]

@Florian, all I can say is that your designs must be very simple if you code once and they work in all browsers from IE6 to Opera. And tell me how you do *automated* testing across IE6, IE7, IE8, Safari, Chrome, Firefox and Opera on a Mac? I can sit down, bash out a few lines in Flex and it runs - yes, across all browsers, with built-in unit testing if I choose. Playing back video in Flash is a train-wreck? I've written a video annotation tool in Flex that works perfectly, thank you very much. Show me your cross-borwser, non-Flash solution then? Are you going to encode all your videos twice (Theora and h.264) and also offer a Flash fallback for IE? Let me know when you've written an app like Aviary in HTML/CSS/JS and I'll buy you a case of beer.

@Carsten, Steve Jobs didn't say that Flash could crash the whole system. That would be an admittance that OSX is a failure if a browser plugin could crash the entire OS. He was talking about figures from crash reports which are sent when programs (in this case Safari) need to be terminated prematurely. There's an argument that this represents an issue with Safari too - Flash can't bring down Chrome on Windows for example. I'm not looking forward to the world where all the crappy programmers are making their ads in HTML5 instead of Flash. Because it's the crappy programmers who are to blame, not Flash.

#5.1 Carsten Schlipf on 02/21/10 at 01:14 AM [Reply]

Hi Darren,

an excerpt from http://www.wired.com/epicenter/2010/01/googles-dont-be-evil-mantra-is-bullshit-adobe-is-lazy-apples-steve-jobs/"

"Whenever a Mac crashes more often than not it’s because of Flash. No one will be using Flash, he says."

Well maybe he hasn't said that and the article does not reflect his words exactly. But as Jobs is quoted here, I understand that it's the system (Mac).

But well, you're right. I doubt that Flash can really bring down the whole system.

#6 venkatnarayan on 04/11/11 at 11:26 AM [Reply]

Great article. Apple is going to fall soon and that will be because of Steve Jobs and his ego.

Add Comment

Name

Homepage

In reply to

HTML-Tags will be converted to Entities.

Standard emoticons like :-) and ;-) are converted to images.

E-Mail addresses will not be displayed and will only be used for E-Mail notifications.

To prevent automated Bots from commentspamming, please enter the string you see in the image below in the appropriate input box. Your comment will only be submitted if the strings match. Please ensure that your browser supports and accepts cookies, or your comment cannot be verified correctly.
CAPTCHA

Enter the string from the spam-prevention image above:

Remember Information? Subscribe to this entry

Occasional Thoughts