HISTORY OF AJAX

Many a year ago, Jeff Rouyer won the Netscape DHTML competition with his flying dude interface and put DHTML on the map. The winning entry, apart from being 3D in design which up until that point had not been seen on the Web, also had something that was quite uncharacteristic of Web pages.
Most of what you see today on the web in terms of DHTML Widgets can be attributed to a small number of people with Jeff leading the way. Granted there has been a progression over the years, but things like scrollers, tooltips, drag and drop, event handling, dynamic clipping etc were pioneered by Jeff many years before they became popular. For example, the sliding doors technique used in CSS today was used around ten years ago with JavaScript by Jeff. In short the guy is the grand daddy of DHTML and everyone else pales in comparison. I kid you not!

One of the unusual features of Jeff’s Award winning site was than rather than have a user traverse across multiple Web pages which consumes resources by opening separate window instances, Jeff was loading content dynamically into an existing interface. In other words all the content was being loaded into an existing element and all other elements were not being changed or swapped. We are talking pre Internet Explorer 4 days here.

Internet Explorer 4 was released a little later to challenge Netscape's dominance on the web, but it did not have a similar method where content could be loaded directly into a layer element by using layer.load();

As the popularity of Internet Explorer 4 gained it became increasingly important to Jeff to maintain cross browser compatibility and initially created a series of document.write() in an external file to allow content to be imported into a pre-existing html file. Obviously this was resource intensive and he went on to improve this method for Internet Explorer 4 by created a little Java Applet to allow Internet Explorer 4 to pass external content into an existing element.

Not long after that, he thought of the idea of using an IFrame technique where content would be loaded into a hidden IFrame and then sucked back out and into a div element. This was the genesis of what today is known as Remote IFrame Scripting. Remember we are talking close to a decade ago which is a good indicator of how far ahead of the curve Jeff was at the time.

Enter Ajax

Recently Garret Smith of Adaptive Path made a good case for using Asynchronous JavaScript and XML and coined the acronym "AJAX" to represent the notion of Remote IFrame Scripting. It is similar to the Remote IFrame technique in that things are handled on the client side, but as we shall see differs in significant ways.

Ajax is more a natural progression from an existing set of techniques as opposed to being something completely brand new. Yet within that progression, things are markedly different than in previous years in a couple of key areas:


· It provides a mechanism to mix and match xml with xhtml.
· It significantly reduces having to continually fetch things from a server (remote IFrame scripting is reasonably comparable here).
· It overcomes some speed bottlenecks that traditional Web development has fallen prey too. In most instances an Ajax based site will load quicker than a comparable traditional Web site.
· When done well, it significantly reduces initial load times.

Before proceeding, it should be noted that Ajax isn't a panacea. It has some problems that it needs to overcome as it continues to mature. But it is a promising field to be involved within and as the years roll on, it probably will greatly affect the way we think about building Web pages and Web applications.

It also needs to be clear that Ajax isn't a technology as such but rather is a technique that combines well with other technologies and techniques. For example, xml, dhtml, css, xhtml. In fact, Ajax really is DHTML with the xmlhttprequest object thrown in. It is amazing to think how one object can change the whole playing field.

As things currently stand in the Ajax field, we almost have two extremes going on. At one end of the continuum we have developers building all sorts of complicated JavaScript tags that in most instances are unnecessary and are overkill. At the other end of the scale we have these little examples floating around the Web that demonstrate how to load xml into a document or do some Ajax form processing.

What is Ajax?

Ajax itself is a technique, but to use the technique effectively one must become familiar with the philosophy behind the technique. In other words, it is not just the use of the technique that is important, but rather developing a different mindset and approach to Web development that is central.

To date, I have mentioned the term Ajax a number of times but have not described the central object that makes Ajax, Ajax. At the heart of Ajax are the xmlhttprequest object and its Microsoft's activeX equivalent. It is this object that allows data to be transferred asynchronously. In case you are not clear what asynchronous means, it is the ability to handle processes independently from other processes. Synchronous which is the opposite of asynchronous, then means that processes are dependent upon other processes. To illustrate let us use a classical Web page scenario.

Assume we have page A and on this page we have a number of elements, including a couple of script and style tags. With synchronous data transfer the script tag needs to be parsed before the next element is parsed. In this way then the next element to be parsed by the browser is dependent upon the script tag being parsed first. Effectively we are creating a bottleneck of one connection between Web page and browser. Style tags and link elements in the head section of the document create the same bottleneck effect. They have to be parsed one at a time before other page elements can download.

Once they have been parsed then the elements in the body section can use concurrent connections to help speed up the download process. For example, most servers handle between 2 to 4 concurrent connections between web page and browser.

Consequently, this means that 2 to 4 images or other page elements can load concurrently. Yet before that process starts what is between the head tags have to be parsed first, which can considerably delay the loading of web pages. Most particularly if you are using multiple CSS and JavaScript tags. Nearly every Web page and blog on the internet uses this method. It isn't hard to see why this may be a problem in terms of increasing page load speeds.

Asynchronously loading methods differ from this scenario because the loading processes are handled independently of each other and to a significant degree overcomes the bottle of traditional Web page design. As will be demonstrated later, the whole gist of this technique is to use minimal JavaScript initially and then push all the head related tags, including the rest of the JavaScript, CSS etc tags through an xmlhttprequest object. This method uses multiple connections rather than the single one that most traditional Web pages use and consequently speeds things up considerably.