Linkbases

One of the most revolutionary features of XLinks is the ability to define links between documents you don't control. For instance, Example 10-1 is an extended link that describes and links three documents neither of the authors of this tutorial has anything to do with. Links between purely remote resources are called third-party links . A third-party link is created when an arc's xlink:from and xlink:to attributes both contain labels for locator elements. Links from a remote resource to a local resource are called inbound links . An inbound link is created when an arc's xlink:from attribute contains the label of a locator element and its xlink:to attribute contains the label of a resource element. Links from a local resource to a remote resource are called outbound links. An outbound link is established when an arc's xlink:from attribute contains the label of a resource element and its xlink:to attribute contains the label of a locator element. Simple links are also outbound links.

An XML document that contains any inbound or third-party links is called a linkbase. A linkbase establishes links from documents other than the linkbase itself, perhaps documents that the author of the linkbase does not own and cannot control. Exactly how a browser or other application will load a linkbase and discover the links there is still an open question. It will probably involve visiting a website that provides the linkbase. When the browser sees the extended link that attempts to establish links from a third website, it should ask the user whether he wishes to accept the suggested links. It might even use the xlink:role and xlink:title attributes to help the user make this decision, though if past experience with cookies, Java applets, and ActiveX controls is any guide, the initial user interfaces are likely to be quite poor and the choices offered quite limited.

Once a browser has loaded a linkbase and arrived at a page that's referenced as the starting resource of one or more of the links in the linkbase, it should make this fact known to the user somehow and give them a means to traverse the link. Once again the user interface for this activity remains to be designed. Perhaps it will be a pop-up window showing the third-party links associated with a page. Or perhaps it will simply embed the links in the page, but use a different color underlining. The user could still activate them in exactly the same way they activate a normal HTML link.

If this is the scheme that's adopted, then it would be useful if the starting resource of the link didn't have to be an entire document, but could rather be just one part of it, such as a specific paragraph, personal name, or tutorial title. Indeed, you can attach an XPointer to the URI identifying the starting resource of the link that chooses a particular part of or point in the starting document. This will be the subject of .