XML-structured documentation has long been a goal of many companies, especially those that sell and support complex products like software and hardware. Having a single source of format-neutral content lets technical writers create content once and then publish it to multiple formats. It also opens up the possibility of content reuse, where a single piece of content can be used in many ways, including manuals, help menus and online knowledge bases.
Despite all its advantages as well as the fact that it directly descended from Standard Generalized Markup Language (SGML), the granddaddy of document technologies, XML has not been widely adopted for documentation.
There are three stages in evolving toward a services-oriented architecture, and if a recent poll by Systinet and WebServices.org is accurate, we're not terribly mature. About one quarter of the 900+ respondents were in phase one, with simple point-to-point services, 17% were in the middle and only 18% were in phase three. 29% said services were "in development" and 9% said they had "no services".
The Darwin Information Typing Architecture (DITA) is finally pushing aside roadblocks that have stood in the way of adopting XML-based publishing. DITA is an XML-based architecture for authoring and producing technical content that was approved as an OASIS standard in May. Although new standards usually gain support quite slowly, DITA is taking off. Software companies are leading the way by using the standard for their own internal needs.
Take Adobe Systems, for example. Tucked into its announcement that FrameMaker 7.2 now supports the DITA Document Type Definition (DTD) was the news that Adobe had used DITA to produce the documentation for Creative Suite 2—all 110,000 pages of it in 14 languages.
What advantages did DITA provide? At a recent conference, Andrew Thomas, senior cross-media producer at Adobe, said DITA tags are easy for users to understand. He added that the standard's topic-orientation lets the company readily share content modules across products. In the past, Adobe relied on manual processes that limited sharing to large, chapter-length blocks of content. Thomas said this inevitably led to "content being rewritten in editing and review" for each new context as well as "higher development and localization costs, longer lead times and lower quality and usability."
Other companies are following Adobe's lead. CAD software vendor Autodesk has migrated its content to DITA, including online help, printed books, tutorials and command references. ATI Technologies, the graphics board maker, purchased the TEXTML server from Ixiasoft to manage its DITA-tagged content. And Sybase recently licensed XMetaL Author DITA Edition for authoring and publishing DITA-encoded content.
So is DITA a panacea for companies producing product-support content? Adopting DITA does not eliminate the necessary first step of modeling content for XML, but it does anticipate the need for custom content models with a method called specialization. Consultants and experienced users admit that although "out-of-the-box" DITA is a great start, many organizations will end up specializing DITA for their own uses. Both Adobe and Autodesk rely on specialized versions of DITA, and other adopters will likely need to do the same.
"DITA's specialization mechanism can be seen as the single most important part of DITA because it allows for the controlled, managed use of XML for technical documentation that enterprises have never had before," says industry veteran Eliot Kimber, senior consulting engineer at Hackensack, N.J.-based Innodata Isogen. "By definition, no generic solution can satisfy all the detailed requirements of any enterprise. At the same time, few but the largest enterprises want to bear the cost of implementing a custom solution entirely from scratch given the existence of generic solutions like DocBook."
For Kimber, the traction DITA is gaining makes perfect sense, as it points right back to the built-in ability to specialize the core DITA tag set. "DITA specialization lets you essentially eat your cake and have it," he says. "You can start with a solid general base (the core DITA types) and, having carefully analyzed your requirements, decide how much specialization to do knowing that your information will continue to be re-usable and interchangeable by all other DITA users."
This makes the initial cost of entry very low but provides a clear path for future refinements that can be developed as needed.
Bill Trippe is president of New Millennium Publishing, a Boston-based consultancy specializing in electronic publishing, content management, SGML and XML.