Building Oracle XML Applications

Handling Doctype Evolution

As time goes on, the DTD of your document collection may change, going through several versions as it evolves. A heterogeneous collection faces that challenge, as well as the challenge of new, unanticipated doctypes entering the system. How can interMedia cope with these changes?

The autosectioner indexes every XML tag, so as soon as a new XML tag appears in your collection, you can use it in WITHIN searches. The only trouble spot to watch out for is tag collision.

The XML sectioner has a harder time handling new tags. With the XML sectioner, the user must define which tags to keep before creating the index. Every undefined tag is thrown away. As new tags appear in the document collection, they are simply thrown away as well.

When this happens, the index owner must modify the index to add the new tags to the index’s list of tags to keep. This is done using an ALTER INDEX command. If our <DamageReport> fragments start to include a <Lawsuit> tag, for instance, the following command:

ALTER INDEX damagereportx REBUILD
PARAMETERS ('add zone section lawsuit tag Lawsuit')

will modify the index metadata and add the tag <Lawsuit> as the section lawsuit. Don’t be scared by the ALTER INDEX . . . REBUILD syntax—the index is not really going to be dropped and recreated. This is a quick metadata change only. In fact, because the index is not dropped and recreated, this will not affect documents already in the system, although documents added after this call with the <Lawsuit> ...