HTML Unleashed. The Emergence of XML: Linking Capabilities of XML

he first part of XML specification fully describes the syntax of the language.  However, there's a second part of the specification that is concerned with linking capabilities of XML documents similar to the hypertext features of HTML.

In some sense, this is beyond the scope of an SGML-like language, as the semantics of elements (including their capability to link) should be left to creators of SGML or XML applications.  On the other hand, links are difficult to implement using common style sheet languages and therefore need support from a deeper level.  For this reason (as well as, apparently, to make XML a better competition to HTML), authors of the language chose to make linking provisions a part of the main body of XML standard.

The linking model of XML (it can be used with SGML as well) is much more versatile than that of HTML.  It is largely based on the efforts of other SGML projects such as HyTime and Text Encoding Initiative (TEI).  I don't describe the detailed syntax here, as it is likely to change in subsequent drafts of the specification.  It is worth noting, however, that all linking constructs are implemented on the level of reserved attributes and their values, not elements.  This means that you can easily turn any element into a link by expanding its list of attributes.

A simple link intended to connect the XML document you're reading to some other Internet resource may be assigned a number of parameters, including the following:

• Strings describing the role of the link in the document and its associated label (for example, this information may be displayed by the browser in its status line);

• A parameter defining whether the linked resource should, upon activating the link, replace the current document, be opened in a new context (for example, in a new window), or be embedded into the current document in place of the link;

• A parameter indicating whether the user may actuate the link or it is activated automatically when encountered by the processing application.

Very important, although perfectly backwards-compatible, enhancements are proposed for the syntax of URLs that are used in XML to address resources.  Using search parts (separated by ?) or fragment identifiers (separated by #) of special form, you can access any part of another XML document without the necessity of attaching a label to it beforehand (such as <A NAME="..."> in HTML).

The syntax of such extended locators in XML is capable of expressing formally such requests as "everything from the third paragraph of the first chapter to the end of the chapter" or "the whole section to which the last sidebar belongs."  Of course "chapters" or "sidebars" are to be declared as corresponding elements in the DTD of the linked document.

Moreover, you may prescribe as to whether the necessary fragment is extracted right on the server where the linked document is stored (which may result in considerable bandwidth savings) or the server sends you the whole document and it's your application that needs to winkle out the part you requested.

XML links can be not only inline but also out-of-line, which means that they do not need to reside in one of the documents connected by the link.  Links can be grouped so that all the links in a group are activated at once.  Finally, you can create collections of interlinked documents and inform the processing application about what other documents contain links to the current one.