Please note that the Tahiti.oracle.com subdomain was decommissioned in 2014. All Oracle documentation is now at docs.oracle.com. However, if you poke around, you can still see database-backed URLs, short URLs with 5-letter book identifiers, and other vestiges of the original technology. This page is archived for historic purposes.
Tahiti was the code name for John Russell's project to make Oracle documentation easier to navigate, filter, and search, built on top of Oracle technologies such as the Database, PL/SQL, and Application Server.
Although the tahiti.oracle.com site which debuted in 2000, was only a single page, the Tahiti infrastructure provided many services for the documentation libraries linked from that page, such as Database, Application Server / Fusion Middleware, and Collaboration Suite / Beehive. As you browsed and searched around these libraries, any page with a /pls/ in the URL was calling on Tahiti technology.
To link to a particular spot in a documentation library, use the URL format http://www.oracle.com/pls/library_code/lookup?id=identifier. Values for library_code include db102, db111, as1014, as1111, cs101, and bee1. The most popular values for identifier are:
- 5-letter all-caps codes representing each book, such as SQLRF for SQL Reference and CNCPT for Database Concepts. (If you do a "View Source" on any doc page, you should see the book code in a meta tag in the head element.)
- A 5-letter book code followed by 3 to 5 digits, representing a topic within a book. There may be none, one, or several of these codes on any doc page. (If you do a "View Source" on the page, you will find any of these target codes as both the name and ID attribute on an anchor tag.)
- An error code such as ORA-00600. (These kinds of lookups require the error message listings to be formatted a certain way, and are primarily found in the Database libraries.)
- The name of a SQL statement or USER_/DBA_/ALL_ or V$ data dictionary view. (Again, primarily in the Database library.)
Different variations of search results can be produced through a REST interface. For example, see Eddie Awad's search plugins for Firefox. Search results can be retrieved in RSS format, by appending the parameter format=rss to the other parameters in a search URL. (Database 10.2 or higher; other libraries TBD.)
The list of books can be retrieved in RSS format, along with summaries and information about creation/last update date, using the URL format /pls/new_html and /pls/new_pdf. The reason for having separate feeds for HTML and PDF is that the likely actions are different in each case. PDFs are easy to download whenever they are updated, in fact some RSS readers can auto-download every linked item. With HTML, you might browse when a new book appears in the library, but the news that the HTML version of a book was updated from revision -02 to -03 is more of an FYI. (Database 11g; other libraries TBD.)