Thoughts on DITA

[Paul Prescod <prescod@gmail.com>]

IMHO, the keys to DITA (relative to DocBook) are the following:

1. Specialization: specialization brings the idea of subtyping to the
document world. DITA's implementation of subtyping is clever in that
it is much simpler than HyTime's and is independent of schema
language. This means that it has a real chance of working well.

Subtyping means that I can configure an editor, formatter, content
management system or whatever to support DITA topics and you can send
me a new kind of topic (perhaps an FAQ). My processor will handle that
FAQ automatically: to the best of its abilities based on subtyping
relationships. If the "best of its abilities" is not good enough then
I can tighten up its configuration. But I always have the choice of
falling back on default behaviours. I don't believe that DocBook has
this feature.

2. Map and Topic orientation: DITA deliverables are created by
combining groups of topics through a map. The map can establish
hierarchical relationships that are analogous to a table of contents
or peer-to-peer relationships that are analogous to the "see also"
lists that in online resources.

3. CONREF: DITA embeds its own variant of the XInclude idea. The
biggest difference between CONREF and XInclude is that XInclude must
either be implemented as a pre-process to DTD/Schema processing or
XInclude elements must be sprinkled throughout your DTD or Schema.
Since ANY is in the content model for XInclude (xi:fallback), XInclude
makes it really easy to violate your DTD or Schema. DITA's Conref is
implemented through an attribute on elements that otherwise have their
normal (strict) content models.

As an implementor of XML authoring solutions, I see each of these as
offering important new opportunities to make the content creation
experience better. They are also each interesting because they offer
more flexibility in how you structure your content at the same time
that they offer more structure around how you take advantage of that
flexibility.

So, for example, specialization makes it easy to customize DITA just
as you customize DocBook. But it constrains you to types of
customizations that allow subtyping to work. You could also emulate
DITA maps and content references with XML entities, but those are not
nearly as structured.

That said, DITA has many flaws and weaknesses when compared to
DocBook.Perhaps DITA is just immature or perhaps it will have gaps
indefinitely. But it seems to me that it has good genes and will
evolve quickly and in interesting ways.

And of course the fourth advantage of DITA over DocBook is hype.

 Paul Prescod