AJ's blog

March 12, 2011

HTML5 – Part III: The Limits

Filed under: HTML5, Silverlight, Software Architecture — ajdotnet @ 5:26 pm

Alternative title: HTML5 – Still HTML, not a RIA platform

The last post was about the chances and benefits of HTML5. But there are also some exaggerated expectations that HTML5 cannot fulfill. Mainly this concerns the term “RIA”, and the effect HTML5 will have in this area.

Or as someone wrote:

“Will HTML 5 one day make Flash, Silverlight and other plug-in technologies obsolete?” (link)

Actually this is a very regular question I have to answer. (Often enough it doesn’t come as question, but as statement. Or accusation. Sic!)

I already detailed how HTML5’s video and canvas will take over tasks that have formerly been solved by other technologies, namely Flash. And some people seem to infer that if Flash is a RIA technology, and HTML5 obsoletes Flash, then HTML5 must obviously make RIA technologies obsolete in general.

Quite wrong. Actually the fact that Flash is used for video and 2D graphics has nothing to do with Flash being a RIA technology. Flash simply happened to be able to delivering these features, both in terms of technical capability as well as broad availability. That it also happens to be a RIA technology is more or less happenstance.

But before moving on we need to clarify the terms “RIA” and “RIA technology”…

My personal definition of RIA technologies relates to the following attributes: Stateful programming model, with some kind of page model, for applications running in a browser sandbox. This includes Flash, JavaFx, Silverlight as a browser plugin (but not in its WP7-platform variation).

Wikipedia applies the terms to Flash, Java (not JavaFx! Sic!), and Silverlight. Sill, this is debatable, and a year ago even Wikipedia had a far broader definition, but in my experience this actually covers the common understanding. Curiously Adobe claims that AIR is their RIA technology, not Flash. But me, Wikipedia, and general consensus agree that Flash indeed is a valid RIA technology.

By the way: Leaving HTML+AJAX out of this picture is by no means meant to be deprecatory, it just reflects common understanding. Wikipedia actually makes the distinction based on the (lack of) necessity to install an additional software framework.

And a final tidbit: Once upon a time Microsoft advertised Silverlight as Flash replacement, addressing video and graphics, just like in the typical Flash use cases. However, even with growing adoption of the Silverlight plugin, Silverlight never became a serious competitor for Flash in that area. (This may actually have played a role in Microsoft’s commitment to HTML5…) Still, Silverlight has long outgrown this narrow definition and later versions has put more emphasis on business features.

So, let’s have a look at where HTML5 reaches its limits and where RIA technologies might kick in. I’ll look at regular web applications before moving on to RIA applications.

Web Applications

HTML5 is going to address standard demands of web applications, including those addressed today by Flash. This will have a crowding-out effect on Flash and RIA technologies in general. But once the demands go beyond being “standard”, RIA technologies will find their niche even in web applications.

On example could be premium video delivery: Some vendors will probably be eager to offer unique selling propositions in the emerging markets of WebTV and high quality HD video content (probably involving DRM).

Since Flash can no longer play the trump of being the only or the broadest available platform in this area, this will also change the picture among the RIA technologies. Especially Silverlight has been very successful in this area recently. Take the Olympic Games or maxdome.

Other examples include complex user interactions that are not feasible with canvas and script, e.g. Mazda’s car configurator, and similarly dynamic data visualizations.

Finally there is the support of additional features. RIA technology vendors certainly have shorter innovation cycles than standards bodies. This especially includes hardware support (camera, game controller, hardware accelerated animations and 3D).

These scenarios all require the user to accept the plugin – which might become a more severe issue if this necessity is less ubiquitous. Thus for the web site provider this always incurs the question whether he can compel his users to use his offering despite that nuisance, or whether he may have to provide a (perhaps simplified and less capable) HTML based version.

RIA Applications

HTML5 won’t turn HTML into a RIA technology. It doesn’t come with a new programming model, doesn’t change server side processing, page model, and postbacks, doesn’t change that fact that the HTML ecosystem really is a conglomerate of diverse technologies.

Many applications – be it multimedia, some kind of hardware dependency, or line of business – simply require the potential of rich client applications. For these, HTML simply cannot deliver what is necessary. Typical demands include:

  • Data centric applications: Large amounts of data; data input with complex validations, lists with extended feature sets, …
  • Usability: Immediate feedback (not possible with postbacks), dynamic UIs with forms built dynamically or changing depending on user input, …
  • Business features: Printing, graphical presentation, Office integration, …
  • Offline capability…
  • Connectivity: Communication patterns such as peer-to-peer, server pushes, …
  • Multimedia and hardware support: Animations, camera, microphone, multitouch, …
  • Rich platform: Stateful programming model, component model, feature rich controls, rich databinding capabilities, …

While these are certainly not the demands for typical web applications. But intranet applications and applications addressing some distinct or closed user group on the web are very well within this category. Prominent example is SAP, one can also think of WebTV portals, home banking, or other.

In the past java applets were often used to cover these demands. Recently AJAX approach have spread, but while this worked to some degree, it often falls short of meeting the demands completely. From a technical perspective, RIA technologies are the adequate choice in these scenarios. And (in my opinion), Microsoft Silverlight is currently the best technology available in that area. Adobe AIR lacks availability and adoption, Flash alone is not sufficient, and JavaFx seems to die a slow death


HTML5 will push RIA technologies out of their makeshift role (video and canvas). However this doesn’t affect the feasibility of employing RIA technologies on their own turf, i.e. beyond-HTML-capability demands in web applications and fully fledged RIA applications.

However, since this “pushing out of RIA technologies” mainly affects Flash, HTML5 has an interesting effect on the RIA market: Broad availability is no longer a strong USP for Flash, which is to the benefit of Silverlight. Add the hazy prospect of JavaFx and the fact that Silverlight is not only a RIA platform, but also enters devices (WP7, WebTV), and HTML5 may actually further the adoption of Silverlight – not as cross-platform tool, as it was once intended, but in all areas not covered by HTML5.

The one argument in favor of HTML5 – which no RIA technology is likely to ever achieve – is its universal availability across all platforms, even if that comes at a cost.

Where are we?

The conclusion of this little series may be as follows: The conflict, or enmity, between HTML5 and RIA that some people see (or exaggerate) doesn’t really exist. There may be a competition between HTML5 and Flash, but even that may turn out differently form what people expect.

Actually HTML5 and RIA complement each other. There are areas in which one technology certainly make more sense than the other, other areas in which there is a choice, again other areas in which a combination of both may work best; even areas in which neither is an ideal choice. E.g.…

  • A web application addressing the broadest available audience? HTML5.
  • An LOB application with high usability demands? Silverlight.
  • A mobile application addressing a broad audience? HTML5. As long as not tighter device integration is necessary, in which case one has to address several mobile OSes…

And between these black-and-white examples there’s a lot gray areas, to be decided on a case-by-case basis. And usually the important thing is not exactly which technology you favor. The important thing is to make an informed decision, aware of the pros and cons, and not solely based on political opinions of certain apologists.

That’s all for now folks,

kick it on DotNetKicks.com

March 10, 2011

HTML5 – Part II: New Standard for Web Applications

Filed under: HTML5, Software Architecture — ajdotnet @ 6:35 pm

As I laid out in the first post, a lot of people talk about HTML5, but there are also a lot of misconceptions – and misguided expectations – about HTML5. So, what is HTML5, anyway? And for whom?

HTML5 is a lot of things for a lot of people. For some it is a vehicle to rid the web of patent laden video formats. For Apple it is a means to keep Flash off iPhone and iPad. Microsoft uses it to push IE9. Some people are using it to – at least try to – bury RIA. Anything else…? Oh yes, some people actually look into HTML5 technologies.

In an attempt to join the last group, here is my opinion on what HTML5 is and what it will mean.

The Standard

Formally, HTML5 is just the next version of HTML. (Period!) In terms of scope however it covers some more ground than “HTML” did before. Included is not only the markup language we know as HTML 4.1, but also the formerly distinct standards XHTML and DOM.

Given that HTML 4.0 became a recommendation in 1997 one can only say (shout!) “It’s about time!”. The fact that HTML 4.x cannot be considered “state of the art” for quite some time now is exactly the reason why browser vendors chose to implement proprietary extensions and why technologies like Flash – even ActiveX – have to fill in the gaps.

HTML5 is available as working draft (latest version today from January 2011) and all major browser venders are more or less committed to implementing it:

Regarding general support for HTML5, the industry for once agrees on something. Surprising enough. Regarding the details however, one has to be careful, which part of HTML5 is supported by which browser to what degree. It may be going to take some time – and more than one version number – until we get broad HTML5 support on all browsers.

For some time the critics gloated at the fact that HTML5 has barely reached “working draft” status and that “recommendation” status is not expected before 2022. Even if that were the case, it would have been totally irrelevant, as the industry is implementing and therefore standardizing parts of HTML5 right now. Additionally the W3C group is just now reconsidering its approach and looking for a more timely “delivery”. TODO

So far for the term, now for the content…

The Details

In the details, HTML5 is a conglomerate of different – and not necessarily related – improvements and new features. Of course this includes the two features mentioned mostly: video and 2D graphics (a.k.a. “canvas”).

Video is nothing spectacular: It simply displays a video, albeit without depending on some 3rd party plugin (namely Flash). Benefit for web developers: a standardized API based on HTML and JavaScript (rather than having to learn another technology). Benefit for the user: He can watch the video, independently on the device and plugin availability.

Canvas allows 2D graphics, even (with JavaScript) animated. This again allows (in principle) to address use cases formerly the domain of Flash. Diagrams for usage statistics, stock exchange rates, and so on. Contrary to simple (server rendered) images this could be including user interaction. This even may include more complex things, like web games; the technology is up to it, as this classic proves.

Regarding multimedia, one probably hast to mention audio support for a complete picture.

Video and canvas are mentioned quite often, probably because they constitute the use cases today most often addressed using Flash. Still, it would be unfair to reduce HTML5 to these two.

If it comes to plain old markup code HTML5 offers some improvements regarding java script control, as well as semantic tags like “header” and “footer”.

Regarding user interaction HTML5 offers new types of input fields (e.g. date picker), complete with validations (e.g. textbox with email address). Also add APIs for drag-and-drop, browser history management, and some other.

You may find this in more detail at the W3C or on Wikipedia. A nicely illustrated introduction can be found here (German, but the pictures shouldn’t need translation ;-)).

Yet this still doesn’t conclude the list. For a complete picture one has to name topics closely related to (even if not formally part of) HTML5: CSS3, Web Storage, and (especially for mobile applications) Geolocation.

It should be noted that the recent improvements in JavaScript execution in all major browsers (well… – IE starting with IE9) also contribute to HTML5: Many of the features unfold their full potential only with scripting. It’s a fair assumption that HTML5 will cause a further growth in scripting in general, and probably the adoption of respective AJAX frameworks.

The Impact

At a closer look HTML5 is a mixture of rather different and unrelated things – none of them especially outstanding. Basically HTML is only upgraded to accommodate current state-of-the-art technologies. All together a (long overdue and therefore a little extensive) evolution, but certainly no revolution or “groundbreaking upgrade”, as some like to think.

Therefor the relevance of HTML5 is not the functionality itself. It stems from two facts:

  1. The broad support from all major browser vendors. No matter why they do it, that fact that they agree on HTML5 in such an unequivocal and joined fashion is without precedence. This will likely ensure that all browsers level on comparable features and capabilities. Which is important to break todays disparities and incompatibilities. HTML5 shows all promises of becoming a platform for the web that is state-of-the-art as well as broadly available. Something HTML has increasingly failed at in recent years.
  2. The timely appearance together with the emerging mobile ecosystem (smart phones, pads). In this area we have a far more diverse and inhomogeneous platform landscape than on desktops (iOS, Android, WP7, other mobile OSes, desktop OS adaptions). No platform has a dominance similar to Windows, thus vendors need to address more than one platform. And web applications built for mobile devices are the only feasible cross platform approach available. Even if HTML5 lacks full device integration (e.g. phone integration or multitouch), it goes a long way with web storage, geolocation, and rudimentary offline capabilities.

To conclude: HTML5 is not going to be some academic standard, it will be a true and broadly supported industry standard. Together with browser improvements, especially regarding script engines, HTML will become an adequate development platform for state-of-the-art web applications and the emerging mobile area.

For web developers HTML5 is a good thing. At least as long as the agreement among the browser vendors holds – and as long as we don’t have to wait another 10 years for the next version.

That’s all for now folks,

kick it on DotNetKicks.com

March 7, 2011

HTML5 – Part I: The Hype

Filed under: HTML5, Silverlight, Software Architecture — ajdotnet @ 5:17 pm

META: This blog got a little sleepy recently. But actually I’ve been spending quite some time writing blog posts, albeit for our company blog. And I’m planning to reuse some of that content here (this post is actually the first one). Also I’ve been busy in several new areas, including WP7, and I may have something to say about those, too. So, this blog is still alive and kicking.

There’s a small hype around HTML5 for some time now. Ironically the reason were more political than for technical ones, since browsers begin to support HTML5 only just now. Be it organizational hassles, discussions about a video format, or a certain company having issues with Flash on their iPlatform. And since Microsoft has announced its decision to make HTML5 their cross platform strategy, the last big browser vendor has joined the camp.

And this is a good thing! Not only homogenizes HTML5 the web platform again, it is also the only feasible platform for cross platform development in the mobile world, which is much more diverse than our desktop eco system.

On the other hand I am regularly irritated about what people think HTML5 will be able to accomplish. Especially in the relation to RIA applications that are all but dead, according to various sources:

“HTML5, a groundbreaking upgrade to the prominent Web presentation specification, could become a game-changer in Web application development, one that might even make obsolete such plug-in-based rich Internet application (RIA) technologies as Adobe Flash, Microsoft Silverlight, and Sun JavaFX.” (link)

“Will HTML 5 one day make Flash, Silverlight and other plug-in technologies obsolete? Most likely, but unfortunately that day is still quite a way off.” (link)

“Sure maybe today, we have to rely on these proprietary browser plugins to deliver content to users, but the real innovative developers and companies are going to standard on HTML 5 and in turn revolutionize how users interact with data.  We all want faster web applications and the only way to deliver this is to use HTML 5.” (link)

To put it bluntly: HTML is no RIA technology and HTML5 is not going to change that. Thus Silverlight is a valid choice for any RIA application today, and it will be one tomorrow.

On the other hand, HTML5 is certainly going to deliver features that today are the domain of RIA technologies, namely Flash. And this will affect RIA technologies in some way.

Then how will HTML5 affect RIA? Well, I’m afraid, it’s not that simple and there is no short and sufficient answer that I’m aware of. In order to decide – probably on a case by case basis – what HTML5 can do, in which use cases HTML5 is the right answer, and in which cases RIA technologies still are the better choices, we need to take a closer look at some details…

To keep it a little more concise, I’m going to break with my usual habit of very long posts. I’m splitting the remainder into two additional not-quite-so-long posts:

Stay tuned,

Create a free website or blog at WordPress.com.