The standard HTML document model is a static one. Once displayed on the browser, a document does not change unless the user initiates some activity like selecting a hyperlink with the mouse. The developers at Netscape Communications found that limitation unacceptable and built in some special features to their Navigator browser that let you change HTML document content dynamically. In fact, they provide for two different mechanisms for dynamic documents, which we describe in detail in this chapter. Internet Explorer supports some of these mechanisms, which we'll discuss as well.
We should mention that many people believe dynamic documents will be obsolete in a very short time, displaced by plug-in browser accessories and, in particular, applets. Nonetheless, Netscape and Internet Explorer continue to support dynamic documents, and we believe the technology has virtues you should be aware of, if not take advantage of, in your HTML documents. [the section called "Applets"]
If you remember from our discussion in Chapter 1, HTML and the World Wide Web, the client browser initiates data flow on the Web by contacting a server with a document request. The server honors the request by downloading the document. The client subsequently displays the document's contents to the user. For normal web documents, a single transaction initiated from the client side is all that is needed to collect and display the document. Once displayed, however, it does not change.
Dynamic documents, on the other hand, are the result of multiple transactions initiated from either or both the server side and the client side. A client-pull document is one that initiates multiple transactions from the client side. When the server is the instigator, the dynamic document is known as a server-push document.
In a client-pull document, special HTML codes tell the client to periodically request and download another document from one or more servers on the network, dynamically updating the display.
Server-push documents also advance the way servers communicate with clients. Normally over the Web, the client stays connected with a server for only as long as it takes to retrieve a single document. With server-push documents, the connection remains open and the server continues to send data periodically to the client, adding to or replacing the previous contents.
Netscape is currently the only browser able to handle server-push dynamic HTML documents correctly; both Internet Explorer and Netscape support client-pull documents. With other browsers, you might see only part of the dynamic document at best. At worst, the browser will completely reject the document. Unfortunately, because dynamic documents are client-server processes, they don't work without an HTTP server. That means you can't develop and test your dynamic HTML documents stored as local files, unless you have a server running locally, as well.
As always, we tell you exactly how to use these exciting but nonstandard features, and we admonish you not to use them unless you have a compelling and overriding reason to do so. We are particularly strident with that admonition for dynamic documents, not only because they aren't part of the HTML standard, but because dynamic documents can hog the network. They require larger, longer downloads than their static counterparts. And they require many more (in the case of client-pull) or longer-term (for server-push) client-server connections. Multiple connections on a single server are limited to a few of the vast millions of Web users at a time. We'd hate to see your readers miss out because you've created a jiggling image in a dynamic HTML document that would otherwise have been an effective and readily accessible static document more people could enjoy.